When autocratic leaders become an option-uncertainty and self-esteem predict implicit leadership preferences

Journal of Personality and Social Psychology

When Autocratic Leaders Become an Option�Uncertaintyand Self-Esteem Predict Implicit Leadership PreferencesChristiane Schoel, Matthias Bluemke, Patrick Mueller, and Dagmar Stahlberg

Online First Publication, May 2, 2011. doi: 10.1037/a0023393

CITATION

Schoel, C., Bluemke, M., Mueller, P., & Stahlberg, D. (2011, May 2). When Autocratic Leaders

Become an Option�Uncertainty and Self-Esteem Predict Implicit Leadership Preferences.

Journal of Personality and Social Psychology. Advance online publication. doi:

10.1037/a0023393

When Autocratic Leaders Become an Option—Uncertainty andSelf-Esteem Predict Implicit Leadership Preferences

Christiane SchoelUniversity of Mannheim

Matthias BluemkeUniversity of Heidelberg

Patrick Mueller and Dagmar StahlbergUniversity of Mannheim

We investigated the impact of uncertainty on leadership preferences and propose that the conjunction of

self-esteem level and stability is an important moderator in this regard. Self-threatening uncertainty is

aversive and activates the motivation to regain control. People with high and stable self-esteem should

be confident of achieving this goal by self-determined amelioration of the situation and should therefore

show a stronger preference for democratic leadership under conditions of uncertainty. By contrast, people

with low and unstable self-esteem should place their trust and hope in the abilities of powerful others,

resulting in a preference for autocratic leadership. Studies 1a and 1b validate explicit and implicit

leadership measures and demonstrate a general prodemocratic default attitude under conditions of

certainty. Studies 2 and 3 reveal a democratic reaction for individuals with stable high self-esteem and

a submissive reaction for individuals with unstable low self-esteem under conditions of uncertainty. In

Study 4, this pattern is cancelled out when individuals evaluate leadership styles from a leader instead

of a follower perspective.

Keywords: uncertainty, leadership style, self-esteem level, self-esteem stability, Implicit Association Test

Groups are characterized by an internal structure that is more or

less hierarchic. Although some groups make shared decisions with

all members having an equal say in the process, in many groups

there are one or a few individuals who take on the leadership role

to achieve collective goals. How the group is led and who is held

accountable to what degree for the group’s decisions depends on

leadership style. A more democratic leadership style is character-

ized by a high degree of group members’ involvement and partic-

ipation, whereas an autocratic leader makes decisions without

asking the group members for their input (e.g., Van Vugt, Jepson,

Hart, & De Cremer, 2004). Leadership research has shown the

effects of different leadership styles with regard to two different

outcome variables: group satisfaction and group productivity.

Group member satisfaction was more strongly associated with

democratic than autocratic leadership; however, results were

mixed with regard to the group’s success (see Bass, 1990). In line

with contingency models of leadership and group effectiveness,

the success of a certain leadership style does not depend only on

the leader per se but also on the situational context (F. E. Fiedler,

1964; Vroom & Yetton, 1973). As normative models, they specify

those situations in which a democratic or autocratic leadership

style should be more successful and, hence, applied.

Our research focuses not on actual, but supposed, leadership

effectiveness. In our view, leadership emergence and behavior are

strongly influenced by the followers’ and leaders’ beliefs about

what constitutes effective leadership in a given situation. Because

strong democratic values prevail under normal circumstances in

most Western societies, democratic leadership is clearly preferred

over autocratic leadership. But when and why do individuals for-

sake democratic participation and prefer an autocratic leadership

style? The object of this article is to shed some light on the specific

factors that motivate people to submit to an autocratic leader.

Specifically, we are interested in the role of uncertainty and the

combined influence of self-esteem level and stability on leadership

preferences.

Leadership Styles

One of the primary perspectives on the study of leadership

emphasizes the behavior of leaders, that is, what leaders do and, in

particular, how they do it to attain group goals. Leadership styles

can differ in how power is distributed, whose needs are met, and

how decisions are made (Bass, 1990; Lewin, Lippitt, & White,

1939; Yukl, 1998). Although different terms such as employee and

production centered (Likert, 1961) or consideration and initiating

structure (e.g., Fleishman, 1973; Stogdill, 1974) are used, there

seems to be a general distinction between two clusters of leader-

ship styles, with democratic and autocratic leadership at the ex-

treme ends of the spectrum (Bass, 1990; Vroom & Yetton, 1973).

Democratic leaders, on the one hand, encourage group members to

Christiane Schoel, Patrick Mueller, and Dagmar Stahlberg, Department

of Social Psychology, School of Social Sciences, University of Mannheim,

Mannheim, Germany; Matthias Bluemke, Psychological Institute, Univer-

sity of Heidelberg, Heidelberg, Germany.

Correspondence concerning this article should be addressed to Chris-

tiane Schoel, Department of Social Psychology, School of Social Sciences,

University of Mannheim, 68131 Mannheim, Germany. E-mail:

[email protected]

Journal of Personality and Social Psychology © 2011 American Psychological Association2011, Vol. ●●, No. ●, 000–000 0022-3514/11/$12.00 DOI: 10.1037/a0023393

1

employ their own methods and policies and elicit equal input when

decisions are to be made. Autocratic leaders, by contrast, dictate

methods and stages of goal attainment and are unconcerned about

followers’ autonomy and personal development (Lippitt, 1940).

The primary difference between democratic and autocratic leader-

ship lies in the degree of participation granted to other people or,

on the other side of the coin, in the amount of control that group

members have over decision-making processes and courses of

action (Bass, 1990; Van Vugt et al., 2004). In the present studies,

we focus on the evaluation of these two opposing leadership styles.

Distinguishing Leadership Evaluations

Autocratic leadership is typically viewed with aversion (Nielsen

& Miller, 1997; Rutte & Wilke, 1985; Samuelson & Messick,

1986; Van Vugt & De Cremer, 1999), yet a number of studies have

reported divergent evaluations depending on the specific dimen-

sion to be judged. Studies contrasting democratic and autocratic

leadership have consistently shown a higher level of group morale

and work satisfaction under democratic leadership, but results have

been mixed with regard to group productivity (Bass, 1990; Lewin

et al., 1939; Miller & Monge, 1986; Neuberger, 1972). Therefore,

we deem it necessary to distinguish followers’ evaluations of

leadership valence (pleasant vs. unpleasant) from their evaluations

of success (success vs. failure; see also Hogg, 2007, p. 717).

Valence judgments can be seen as the affective, and success

ratings as the cognitive, attitude component of leadership evalua-

tions (multicomponent model of attitudes; Eagly & Chaiken,

1993). These two dimensions do not necessarily coincide. A well-

liked leader might be seen as quite ineffective, and a disliked

leader might be seen as very successful. Despite the differences

between satisfaction and productivity (i.e., valence and success),

the previous findings indicate a general aversion to autocratic

leadership. Therefore, we argue that the default attitude is

prodemocratic. In particular situations, however, we posit that

autocratic leadership becomes an option and that it is crucial to

differentiate between valence and success. Specifically, we inves-

tigated the way in which leader and follower characteristics relate

to perceived uncertainty.

Leader Characteristics and Situational Aspects

Normative leadership theories make predictions about which

leadership style is most effective in a given situation. For instance,

in his contingency theory, F. E. Fiedler (1964, 1967) argued that

autocratic leadership would be most effective in situations of either

very low or very high control. A low control situation would arise

when the structure of the task is unclear, the leader’s position

power is low, and/or leader–member relations are poor. The op-

posite is the case in situations of extremely high control. Demo-

cratic leadership, by contrast, should be most effective in contexts

where situational control lies in between these two extremes. In

their normative decision theory, Vroom and Yetton (1973) have

proposed that autocratic leadership is only effective in decision-

making situations when the task has a clear structure and when

follower commitment and support is high; otherwise, a democratic

leadership style should be more successful. Thus, both models

make different predictions about which leadership style—

democratic versus autocratic—is successful in an ambiguous sit-

uation. Whereas both normative theories predict objective out-

comes as dependent variables, we focus on subjective leadership

evaluations and argue that a person–situation interaction deter-

mines the preferred leadership style. Thus, a particular leadership

style may be seen as more successful, even if it is not.

Follower Characteristics and Situational Aspects

Empirical evidence that follower’s leadership attitudes are in-

fluenced by threatening situations and personality characteristics

comes from the authoritarianism literature and the terror manage-

ment approach. Submissiveness to an autocratic leader is related to

authoritarian submission (e.g., Altemeyer, 1996). Although there is

a long tradition of explaining authoritarianism by interindividual

differences (Adorno, Frenkel-Brunswik, Levinson, & Sanford,

1950; Allport, 1954; Rokeach, 1954; G. D. Wilson, 1973), empir-

ical evidence suggests that authoritarian attitudes and related be-

haviors are also induced by situational factors such as threats and

personal insecurity (e.g., Doty, Peterson, & Winter, 1991; Duckitt

& Fisher, 2003; Sales, 1972, 1973; Sales & Friend, 1973). For

instance, McCann (1997) found that during years of high threat,

presidential elections were more influenced by the perceived

strength and power of the candidates than during times of low

threat. Furthermore, there is support for the interplay of situational

threat and individual predispositions such as preauthoritarianism

(e.g., Feldman, 2003; Feldman & Stenner, 1997; Lavine et al.,

1999; McCann, 2008; Rickert, 1998; Stenner, 2005).

Taken to the extreme, terror management theory supposes that

people who are reminded of death or who perceive a deadly threat

(mortality salience) are the most likely to favor charismatic and

autocratic over relationship-oriented democratic leaders (e.g., Co-

hen, Solomon, Maxfield, Pyszczynski, & Greenberg, 2004). In-

creased support for former charismatic U.S. President George W.

Bush and his counterterrorism policies has been consistently found

under mortality salience and reminders of 9/11 (Cohen, Ogilvie,

Solomon, Greenberg, & Pyszczynski, 2005; Landau et al., 2004;

Nail & McGregor, 2009). More recently, individual difference

variables have also been found to moderate the link between

mortality salience and leadership preferences (Gillath & Hart,

2010; Kosloff, Greenberg, Weise, & Solomon, 2010; Vail, Arndt,

Motyl, & Pyszczynski, 2009; Weise et al., 2008).

Uncertainty and Its Reduction

One commonality that all these different situations share is that

they are ambiguous, unpredictable, and/or associated with loss of

control and potential harm, and thus, they may have a direct impact

on the individual by inducing personal feelings of uncertainty

(Kagan, 1972). Building on social identity theory, Hogg (2009)

argued that self-uncertainty arises from threats to one’s self-view

or to interpersonal relationships (personal self) on the one hand

and threats to one’s membership in social groups (social self) on

the other. We follow this broad definition. The uncertainty manip-

ulations we apply in our studies ask participants to think of

self-threatening uncertainty, but we leave it to the individual

whether he/she thinks of threats to his/her personal or social self.

Although uncertainty can be perceived as a challenge (e.g., Sor-

rentino, Short, & Raynor, 1984; T. D. Wilson, Centerbar, Kermer,

& Gilbert, 2005), we investigate uncertainty as a threat to the self

2 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

(Loseman, Miedema, van den Bos, & Vermunt, 2009). Thus,

whether understood as a persistent individual trait (e.g., De Cremer

& Sedikides, 2005; Greco & Roger, 2001) or as a state induced by

contextual variations (van den Bos, 2001), uncertainty should be

extremely aversive and accompanied by feelings of reduced con-

trol over one’s life (e.g., Hogg, 2000; Sedikides, De Cremer, Hart,

& Brebels, 2010).

Consequently, reducing uncertainty has been claimed to be

among the most basic of human needs (e.g., Kagan, 1972; Maslow,

1954). Uncertainty can be reduced, for instance, by experiencing

social integration and group identification (Hogg, 2000), perceiv-

ing fairness (van den Bos & Lind, 2002), or establishing certainty

in unaffected domains such as political attitudes (McGregor,

Zanna, Holmes, & Spencer, 2001). Weary and Edwards (1996; see

also Weary, Tobin, & Edwards, 2010) argued that the choice of

coping strategy to overcome uncertainty depends on a deliberative

process of weighing success expectancies of different potential

resolution strategies. Whether a strategy is employed depends on

the expectation that it will be most successful in reducing uncer-

tainty. When the perceived uncertainty is self-relevant, most indi-

viduals will value uncertainty reduction. Individual differences,

however, should exist with regard to the expectancies associated

with uncertainty reduction strategies. In the following sections, we

explore the role of self-esteem level and self-esteem stability for

success expectancies, before we tie self-esteem back to evaluations

of democratic and autocratic leadership.

Level of Self-Esteem

The expectation of successfully performing an action plan

should be especially high (a) when a person is self-confident about

his/her performance of the behavior and (b) when a person is at the

same time optimistic that this behavior leads to the desired out-

come of reduced uncertainty. The crucial dispositional variable

that determines both components is level of self-esteem. Individ-

uals high in self-esteem (high SEs) have been found to hold a

positive self-view (Baumeister, Campbell, Krueger, & Vohs, 2003;

Ross & Wilson, 2002; A. E. Wilson & Ross, 2001), to believe they

are superior to others in many domains (Brown, 1986; Marsh,

1986), to be self-confident in their viewpoints and actions, and,

thus, to be willing to voice their opinion (Brockner et al., 1998;

LePine & Van Dyne, 1998). In contrast, individuals low in self-

esteem (low SEs) hold more negative attitudes toward themselves

(Baumeister, 1993; Rosenberg, 1979) and are less self-confident

(Brockner, 1983), especially after failure (McFarlin & Blascovich,

1981). In addition, high SEs have been shown to be more opti-

mistic (e.g., Scheier, Carver, & Bridges, 1994) and to hold more

positive illusions than low SEs (Aspinwall & Taylor, 1992;

Brown, Collins, & Schmidt, 1988; Taylor & Brown, 1988). There-

fore, high SEs should be more self-confident and optimistic about

successfully reducing uncertainty than low SEs, who should have

more doubts that their behavior would actually make a difference

in overcoming an uncertain situation.

Stability of Self-Esteem

In addition to absolute level of self-esteem, individuals differ in

the stability of their self-esteem (for an overview, see Kernis,

2005; Kernis & Lakey, 2010). Individuals with stable self-esteem

(stable SEs) have secure and enduring feelings of self-worth and a

clearly defined self-concept. They rely on internal more than on

external validation strategies, and their self-worth is not much

affected by situational short-term variations. Individuals with un-

stable self-esteem (unstable SEs), however, hold unsettled and

insecure feelings of self-worth, and their self-concept is character-

ized by confusion, conflict, and self-doubt. They experience sub-

stantial changes of self-esteem level in the presence of external

evaluative cues (e.g., interpersonal feedback) and internal self-

generated information (e.g., salience of specific self-aspects). Not

surprisingly, self-esteem stability affects the self-regulatory styles

of goal pursuit (Deci & Ryan, 1985). Whereas stable SEs are

agentic and self-determined, unstable SEs are more influenced by

external and internal pressures (Kernis, Paradise, Whitaker,

Wheatman, & Goldman, 2000). This pattern should extend to

uncertainty reduction, too: Stable SEs should be more self-

confident and self-determined when it comes to successfully re-

ducing uncertainty than unstable SEs, who should be more likely

to rely on others.

Impact of Level and Stability of Self-Esteemon Choice of Action Plan

During decision making, especially in (uncertain) group situa-

tions, one can generally try to maximize or minimize one’s input.

A democratic leadership style encourages greater input by follow-

ers than an autocratic leadership style. Whether people prefer

increasing or reducing their input should depend on the expecta-

tion that their chosen strategy is potentially successful in reducing

uncertainty. As high SEs are more confident and optimistic, their

process of weighing the alternative action plans should result in

higher expectations of successfully performing those behaviors

that make a difference in the situation, whereas low SEs would not

believe that they could successfully carry out action plans that

reduce uncertainty. In a comparable process, stable SEs should

prefer an agentic and self-determined way of uncertainty reduc-

tion, whereas unstable SEs should be more inclined to rely on

external sources. Therefore, both high and stable SEs should be

motivated to maximize their input, whereas low and unstable SEs

should prefer to minimize their input in decision-making situations

under uncertainty. Given the differences in self-concepts, stable

high SEs should strive for even more participation and input than

they normally would. This, in turn, is compatible only with dem-

ocratic leadership. Hence, their preference for democratic leader-

ship should be increased. We refer to this effect as the democratic

reaction. By contrast, unstable low SEs should place the respon-

sibility for the decision-making process on someone else. Not

trusting the effectiveness of raising their own voice, they look for

an autocratic leader to reduce uncertainty. Hence, their evaluation

of the autocratic leadership style will become more favorable. We

refer to this effect as the submissive reaction. As we assume that

level and stability of self-esteem operate in the same way, we

predict additive effects of the two variables on uncertainty reduc-

tion. This idea is in line with Kernis et al. (2000) who found

additive effects of SE level and SE stability on self-regulatory

styles and self-concept clarity.

3UNCERTAINTY AND LEADERSHIP PREFERENCES

Hypotheses on Valence and Success Evaluationsof Leadership

To develop a full-blown set of hypotheses, we return to the

question of how the different patterns of democratic and submis-

sive reactions can be observed—given that feelings and thoughts

do not necessarily coincide, and given that leadership valence and

leadership success are independent evaluation dimensions. Dem-

ocratic leadership communicates both appreciation and mutual

respect (Tyler & Lind, 1992), whereas autocratic leadership is

linked to strong negative emotions (Bass, 1990; De Cremer, 2007;

Judge, Piccolo, & Ilies, 2004). Therefore, we assume that stable

high SEs and unstable low SEs consensually like democratic

leadership more than autocratic leadership—regardless of the de-

gree of uncertainty. Yet, only stable high SEs should believe that

they can successfully reduce uncertainty by engaging in the prob-

lem solving process. As the outcomes of democratic leadership

depend on the group members’ input, stable high SEs should

associate democratic leadership with success, which is consistent

with their self-view. Such a link, however, is inconsistent with the

self-concept of unstable low SEs. They should expect uncertainty

reduction not from democratic, but from autocratic leaders who (to

them and only to them) promise a means of overcoming states of

uncertainty.

On the basis of these arguments, we claim that under conditions

of certainty, a prodemocratic default attitude prevails (Hypothesis

1). Conditions of uncertainty, however, should cause a democratic

reaction in individuals with high and stable self-esteem (Hypoth-

esis 2) and a submissive reaction in individuals with low and

unstable self-esteem (Hypothesis 3). Both reactions should be

evident mainly on cognitive measures of success but less so on

affective measures of valence. To unravel the differential reac-

tions, we developed explicit and implicit measures that discrimi-

nate valence from success evaluations. Because of strong demo-

cratic values, the expression of preferences for an autocratic leader

might conflict with motivations of social desirability. We therefore

expected the differences between stable high and unstable low

SEs—especially the submissive reaction—to be more pronounced

on implicit than on explicit measures.

Study Overview

Studies 1a and 1b were designed to test Hypothesis 1—that

under conditions of certainty, democratic leadership is clearly

preferred to autocratic leadership—and to validate the new mea-

sures of leadership evaluation. In Study 2, state (un)certainty was

manipulated to provide evidence for Hypotheses 2 and 3, that

stable high SEs show a democratic reaction and that unstable low

SEs show a submissive reaction to uncertainty. Study 3 aimed at

replicating the submissive reaction of unstable low SEs by merely

making trait uncertainty salient. Whereas in Studies 1a–3 we

focused on follower’s reactions to uncertainty, in Study 4, we also

investigated the influence of being in a leadership position com-

pared with being in a follower position.

Study 1a

To assess the initial, unadulterated valence and success associ-

ations linked to democratic and autocratic leadership, in addition to

explicit endorsement of leadership styles, we developed two Im-

plicit Association Tests (IATs; Greenwald, McGhee, & Schwartz,

1998): a valence IAT and a success IAT. Study 1a tested Hypoth-

esis 1 that under conditions of certainty individuals clearly prefer

democratic over autocratic leadership, and that this holds for both

valence and success evaluations as well as for implicit and explicit

measures.

Method

Pretest. To choose the stimuli for the valence and the success

IATs, we conducted a pretest with a set of potential leadership,

valence, and success items. The aim was to reduce the influence of

irrelevant associations between target and attribute dimensions as

far as possible (Bluemke & Friese, 2006; Steffens & Plewe, 2001).

Participants rated 70 leadership items on 7-point scales as more

authoritarian or more democratic (n1 5 16), pleasant or unpleasant

(n2 5 16), or associated with success or failure (n3 5 16). Note

that we used the word “authoritarian” instead of “autocratic,” as it

is much more common in German everyday speech. The adjectives

chosen for the IAT, however, describe an autocratic leadership

style.

We chose five autocratic words (e.g., dominant, directive), five

democratic words (e.g., liberal, team player), five pleasant words

(e.g., diamond, angel), five unpleasant words (e.g., dirt, slime),

five success-associated words (e.g., benefit, victory), and five

failure-associated words (e.g., loss, flop)—see the Appendix for a

complete list of German IAT stimuli and their English translations.

We selected items in such a way that average word length and

number of syllables of complementary target categories and com-

plementary attribute categories were kept constant. The aim was

that autocratic and democratic leadership items (a) could be clearly

distinguished (1 5 authoritarian, 7 5 democratic), Ms 5 2.68

versus 5.23, t(4) 5 –7.36, p 5 .002; (b) were equally strongly

related to positive and negative valence (1 5 pleasant, 7 5

unpleasant), Ms 5 3.97 versus 3.20, t(4) 5 1.38, p 5 .24; and (c)

were equally strongly related to success versus failure (1 5 suc-

cess, 7 5 failure), Ms 5 3.43 versus 3.48, t(4) 5 –0.08, p 5 .94.

In addition, all participants rated 42 valence and 40 success items

presented in mixed order on 7-point scales as more typical for

autocratic or democratic leadership (1 5 authoritarian, 7 5 dem-

ocratic). The tendency to evaluate democratic leadership more

positively than autocratic leadership prevented a full elimination of

cross-category associations for the attribute items: Positive (neg-

ative) items were perceived to be more in line with the democratic

(autocratic) side of the scale, Ms 5 4.45 versus 3.46, t(4) 5 6.15,

p 5 .004, just as success rather than failure items were related to

democratic leadership, Ms 5 4.50 versus 3.47, t(4) 5 4.37, p 5

.01. Although both tendencies already indicated a preexisting

preference for democratic leadership and an aversion to autocratic

leadership, we reduced these asymmetries as much as possible.

Participants and procedure. Forty students from a German

university (20 women; Mage 5 22.2 years) participated in a 15-min

study on “Leadership Attitudes” in return for 1 Euro and a choc-

olate bar. Up to six participants at a time were present in the

laboratory. After participants agreed to informed consent and

provided standard demographic information, we induced state

certainty via a priming procedure adapted from van den Bos

(2001) by asking participants to respond to the following two

4 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

questions: “Please briefly describe the emotions that the thought of

you being certain of yourself arouses in you,” and “Please write

down what you think physically happens to you as you feel certain

of yourself.” Next, participants completed both the valence and the

success IATs in counterbalanced order. Finally, we assessed par-

ticipants’ explicit leadership valence and success evaluations.

Measures.Implicit leadership measures. Both valence and success IATs

required participants to categorize democratic and autocratic target

items jointly with either (a) pleasant and unpleasant or (b) success

and failure attribute stimuli as quickly and as accurately as possi-

ble (see Greenwald et al., 1998, for details). Aside from training

trials for the single discrimination of target and attribute items,

each IAT had two critical blocks. In the compatible block (i.e.,

consistent with democratic values), the same response key was

used for democratic and pleasant (success) stimuli, and the other

key was used for autocratic and unpleasant (failure) stimuli. In the

incompatible block (i.e., inconsistent with democratic values), key

assignments were switched so that responses to autocratic and

pleasant (success) stimuli shared a key, and democratic and un-

pleasant (failure) stimuli likewise.

The IATs were administered according to a seven-block struc-

ture (Greenwald, Nosek, & Banaji, 2003). For each IAT, partici-

pants worked on the following tasks: training the categorization of

leadership items (Block 1: 10 trials); training the sorting of valence

(success) attribute items (Block 2: 10 trials); practicing combined

trials with democratic 1 pleasant (success) stimuli and auto-

cratic 1 unpleasant (failure) stimuli sharing a response key (Block

3: 20 trials); the critical trials of the same combination (Block 4: 40

trials); practicing the reversed response-key assignment for lead-

ership stimuli (Block 5: 10 trials); practicing the incompatible

combination of autocratic 1 pleasant (success) and democratic 1

unpleasant (failure) stimuli (Block 6: 20 trials); and repeating these

trials as a critical test (Block 7: 40 trials). In the critical blocks,

each stimulus was presented twice, and target and attribute stimuli

alternated. IAT block orders were counterbalanced: Half of the

participants completed the compatible blocks first, the other half

started with the incompatible blocks.

IAT data preparation. Mean response latencies of correct

responses (in milliseconds) for each block as well as the proportion

of error rates were available. We excluded those participants

whose results indicated a lack of motivation as evident in error

rates of more than 30% in either the compatible or the incompat-

ible block (two participants for the valence IAT, three participants

for the success IAT). Prior to aggregation, IAT response latencies

lower than 300 ms or higher than 3,000 ms were recoded as 300 ms

and 3,000 ms, respectively. We dropped the first trial of each block

because they typically far exceeded the mean. Latencies for both

IATs were individually z-transformed by subtracting a partici-

pant’s latency grand mean and dividing the result by a participant’s

overall standard deviation (Bluemke & Friese, 2006). This proce-

dure of data trimming and preparation, which is similar to the

D-algorithm (Greenwald et al., 2003), was employed in all of the

following IAT studies.1

The crucial dependent variable, the IAT effect, was calculated

as the mean latency difference between the incompatible and

compatible conditions (Blocks 4 and 7, respectively). Whereas

positive difference scores reflected a tendency to more quickly

associate pleasantness (success) with democratic than with auto-

cratic leadership, negative scores can be interpreted as a stronger

association of pleasantness (success) with autocratic leadership.

The resulting valence and success IAT effects were reliable (as 5

.74 and .80, respectively).

General explicit leadership measures. The general valence

scale asked 3 (3 2) questions, “How positive/pleasant/enjoyable

do you think democratic (authoritarian) leadership is?” (1 5 not at

all, 7 5 extremely). Analogously, the general success scale asked

“How successful/efficient/profitable do you think democratic (au-

thoritarian) leadership is?” (1 5 not at all, 7 5 extremely). The

four subscales each consisted of three questions: general valence

of democratic leadership (a 5 .85) and autocratic leadership (a 5

.91) as well as general success of democratic leadership (a 5 .76)

and autocratic leadership (a 5 .75). To compare the explicit

evaluations with the IAT effects, a difference score of relative

favoritism was computed by subtracting autocratic from demo-

cratic scores such that positive values indicated a preference for

democratic leadership, and negative values indicated a preference

for autocratic leadership.

Results and Discussion

First, both the valence and success IAT effects were subjected to

2 (order of IATs: valence–success vs. success–valence) 3 2 (order

of blocks: compatible–incompatible vs. incompatible–compatible)

analyses of variance. As there were no significant main or inter-

action effects (all ps . .10), we conducted the further analyses

regardless of these order conditions. In line with Hypothesis 1,

two-sided t tests revealed that the valence and success IAT effects

were positive (M 5 0.44, SD 5 0.39, and M 5 0.38, SD 5 0.48,

respectively) and differed significantly from 0, t(37) 5 6.82, p ,

.001, and t(36) 5 4.84, p , .001, respectively, indicating prefer-

ences for democratic compared with autocratic leadership on both

the valence and the success dimensions.

The comparative explicit measures of general valence difference

(M 5 4.15, SD 5 1.75) and general success difference (M 5 1.25,

SD 5 2.34) mirrored these findings by converging on an explicit

democratic preference, t(38) 5 14.86, p , .001, and t(38) 5 3.33,

p 5 .002,2 respectively.

In sum, Study 1a introduced newly developed explicit and

implicit measures to assess valence and success evaluations of

democratic and autocratic leadership. As predicted, when primed

with certainty, participants clearly preferred democratic over au-

tocratic leadership on both explicit and implicit measures as well

as with regard to valence and success. Irrespective of the order of

IAT blocks, and independent from the order of measures, the

average default attitude was democratic (Hypothesis 1).

1 All of the reported results are also significant at least at p , .05 when

analyzed with a D5-algorithm (Greenwald et al., 2003), except for Study 3,

which yielded only a marginally significant beta-weight of .30 ( p 5 .09)

instead of .46 ( p 5 .01) for the interaction term. As the D-algorithms apply

error penalties and our experimental manipulations may have affected the

number or causes of erroneous trials, the D-algorithms run the risk of

introducing nuisance variance that does not reflect the associations prop-

erly, thus masking the hypothesized outcomes.2 Lower degrees of freedom stem from the fact that one participant failed

to complete the explicit measures.

5UNCERTAINTY AND LEADERSHIP PREFERENCES

Study 1b

Study 1b further validated the explicit and implicit measures of

leadership evaluation by examining their intercorrelations and

their links to related constructs such as political orientation (Jost et

al., 2007), right-wing authoritarianism (RWA; Altemeyer, 1996),

and social dominance orientation (SDO; Pratto, Sidanius, Stall-

worth, & Malle, 1994).

Method

Participants and procedure. One-hundred and fifty partic-

ipants (108 women; Mage 5 25.2 years), mainly students from

German universities, completed the 20-min online survey on “New

Measures of Attitudes and Personality.” Participants were re-

cruited through an online pool. In return, they received the chance

to win one of 10 book coupons (worth 10 Euros each). Study 1b

had a 2 (order of measures: explicit–implicit vs. implicit–

explicit) 3 2 (order of content: valence–success vs. success–

valence)3 factorial design, with participants randomly assigned to

conditions. After agreeing to informed consent, participants pro-

vided standard demographic information and stated their political

orientation on a rating scale (1 5 extremely liberal, 9 5 extremely

conservative; cf. Jost et al., 2007). The mean political orientation

for our sample, M 5 3.71 (SD 5 1.50), indicated a slightly more

liberal than conservative ideological position, if one takes the

midpoint of the scale as a reference point, t(149) 5 –10.59, p ,

.001. Next, participants completed the explicit and implicit mea-

sures assessing their attitudes toward democratic and autocratic

leadership in counterbalanced order. Within the block of implicit

and explicit measures, the valence and success IATs were also

balanced. Finally, the RWA and the SDO scales were assessed.

Measures.Implicit leadership measures. Participants completed the va-

lence IAT (a 5 .72) and the success IAT (a 5 .63) as described

in Study 1a in counterbalanced order. Diverging from Study 1a,

however, all participants completed the compatible blocks before

the incompatible blocks to reduce irrelevant error variance that

could obscure correlations.4

Item-based explicit leadership measures. As in the pretest

sample, participants evaluated each of the 10 democratic and

autocratic items used in the valence and the success IATs. Regard-

ing the valence dimension, participants were asked “How pleasant

or unpleasant do you think the following leadership characteristics

are?” (1 5 extremely pleasant, 7 5 extremely unpleasant). Re-

garding the success dimension, the question was “How strongly

are the following leadership characteristics connected to success or

failure?” (1 5 extreme success, 7 5 extreme failure). The ratings

for these items were aggregated to form four different subscales:

valence of democratic (a 5 .60) and autocratic (a 5 .72) items as

well as success attributed to democratic (a 5 .50) and autocratic

(a 5 .71) items. To compare the item-based evaluations with the

IAT effects, items were reverse coded, and a difference measure of

relative favoritism was computed by subtracting autocratic from

democratic item scores such that positive values indicated more

ascribed pleasantness or success for democratic leadership, and

negative values more ascribed pleasantness or success for auto-

cratic leadership.

Additionally, participants were asked to rate the 10 items on

bipolar scales to reflect the relative evaluation as it is prominent in

the IAT: “How democratic or authoritarian do you think the

following leadership characteristics are?” (1 5 extremely demo-

cratic, 7 5 extremely authoritarian; as 5 .54 and .61 for demo-

cratic and autocratic items, respectively).

General explicit leadership measures. General explicit va-

lence and success evaluations of democratic and autocratic lead-

ership were assessed with the four subscales introduced in Study

1a: general valence of democratic leadership (a 5 .93) and auto-

cratic leadership (a 5 .91) as well as general success of demo-

cratic leadership (a 5 .87) and autocratic leadership (a 5 .84).

Difference scores of relative preference were computed by sub-

tracting autocratic from democratic scores.

RWA. Subsequently, participants completed a German trans-

lation of Duckitt, Bizumic, Krauss, and Heled’s (2010) 36-item

scale on RWA (a 5 .92). Sample items were as follows: “It is

important that we preserve our traditional values and moral stan-

dards” (traditionalism, a 5 .82); “The real keys to the ‘good life’

are respect for authority and obedience to those who are in charge”

(authoritarian submission, a 5 .85); and “What our country really

needs is a tough, harsh dose of law and order” (authoritarian

aggression, a 5 .88). Responses were provided on 7-point rating

scales (1 5 strongly disagree, 7 5 strongly agree).

SDO. Finally, participants completed a German 14-item ver-

sion of the SDO scale (Pratto et al., 1994). A sample item (a 5

.87) was as follows: “Some groups of people are simply inferior to

other groups.” Responses were provided on 7-point rating scales

(1 5 strongly disagree, 7 5 strongly agree).

Results

Democratic default attitude. As in Study 1a, the IAT effects

were positive and in line with a democratic default attitude. With-

out counterbalancing the IAT block orders, however, the overall

positive mean scores of the IATs (cf. Table 1) are ambiguous and

should not be interpreted as clearly democratic or autocratic, as

block order was confounded with block compatibility, and mere

learning curves or fatigue effects could have influenced the out-

comes. Therefore, we focus on the results of the explicit leadership

measures.

As intended, two-sided t tests showed that democratic items

were actually seen as more democratic, and autocratic items were

seen as more autocratic, that is, sample means of democratic items

were all significantly lower than the scale midpoint of 4 (all ps ,

.001), and means of autocratic items were all significantly higher

than 4 (all ps , .001). Notably, the item-based valence difference

was significantly higher than 0, indicating a preference for dem-

ocratic leadership, t(149) 5 13.84, p , .001. The item-based

success difference just failed to reach a conventional level of

significance, t(149) 5 1.49, p 5 .13, but pointed in the same

direction. The comparative measures of general valence difference

and general success difference clearly reflected the democratic

3 Comparisons on Fischer’s r-to-z transformations showed no significant

differences between correlations because of order of implicit and explicit

measures, or valence and success measures. Therefore, only combined

correlations are reported.4 We applied this standardized block order in all of the following studies

(Studies 1b–4).

6 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

default attitude, t(149) 5 19.06, p , .001, and t(149) 5 4.72, p ,

.001, respectively (cf. Table 1).

Correlations of explicit and implicit leadership measures.Although the valence and the success IATs were weakly related,

we found support for a specific relationship between implicit and

explicit leadership measures (cf. Table 1). First, the valence IAT

was specifically related to the item-based and the general valence

difference, but not to the corresponding explicit success measures.

Likewise, the success IAT correlated significantly with the item-

based and the general success difference scores, but not with the

corresponding explicit valence measures. This specific pattern of

implicit–explicit correlations indicates that both IATs tap into

different dimensions: valence and success.

Correlations with other explicit measures. No relationships

existed between the valence IAT and the RWA scale or one of its

subscales. The success IAT, however, was negatively related to the

overall RWA scale, driven mainly by the subscale of traditional-

ism, but not by authoritarian submission and aggression. Neither

the valence nor the success IAT correlated with SDO, which lends

support to the unique explanatory value of SDO. Item-based and

general valence differences showed negative correlations with

political orientation, the full RWA scale (including all of its

subscales), and the SDO scale. Item-based and general success

differences mainly replicated these findings, with the exceptions

that both explicit success measures were unrelated to political

orientation, and that the item-based success difference yielded no

significant correlation with RWA-traditionalism.

Thus, the correlations of implicit and explicit measures were

specific, whereas all explicit measures were highly interrelated and

did not show a distinct pattern (cf. Table 1).

Discussion

Study 1b replicated the finding that democratic leadership is

preferred to autocratic leadership on explicit measures and re-

vealed specific correlations between explicit and implicit mea-

sures. The size of the implicit–explicit correlations is in line with

previous meta-analyses across various domains (Greenwald,

Poehlman, Uhlmann, & Banaji, 2009; Hofmann, Gawronski,

Gschwendner, Le, & Schmitt, 2005; Nosek, Greenwald, & Banaji,

2005). Importantly, both IATs displayed characteristic correlations

with their respective explicit leadership evaluations. In addition,

the success IAT effects were significantly, and negatively, related

to RWA, and in particular to RWA-traditionalism. These findings

corroborate how important it is to distinguish between success

expectations and valence attributions associated with leadership.

That RWA-submission and RWA-aggression failed to turn up

similarly strong relationships may be due to stronger social desir-

ability effects, resulting in more pronounced implicit–explicit dis-

crepancies (Degner, Wentura, & Rothermund, 2006). Whereas

these results indicate that the success IAT displays convergent

validity, negligible relationships to SDO indicate divergent valid-

ity for the implicit measures, too. The new explicit leadership

scales, both item-based and general evaluations of leadership va-

lence and success, correlated less specifically with preexisting

dimensions such as political orientation, RWA and SDO, attesting

to the general conservative belief and value system as an under-

lying dimension.

Study 2

Whereas Study 1a demonstrated that, under certainty, most of

the participants clearly preferred democratic over autocratic

leadership, Study 2 aimed at providing empirical evidence for

the hypothesized difference in reactions to uncertainty of stable

high and unstable low SEs. We predicted that, compared with a

certainty condition, state uncertainty would increase stable high

SEs’ success expectations of democratic leadership because of

their confidence in overcoming the perceived uncertainty re-

sulting in a democratic reaction (Hypothesis 2). In contrast,

state uncertainty should reduce unstable low SEs’ success eval-

uations of democratic leadership because of doubts regarding

their own ability to amend the situation, resulting in a submis-

sive reaction (Hypothesis 3).

In other words, we postulated that level and stability of self-

esteem function conjointly in predicting leadership evaluations

under uncertainty. Consistently, in a recent meta-analysis, Okada

(2010) found that self-esteem level and instability were not inde-

Table 1

Descriptions and Intercorrelations in Study 1b

Dependent variable M SD 1 2 3 4 5 6 7 8 9 10 11 12

1. Political orientation 3.71 1.50 — .26pp .45pp .38pp .39pp .37pp 2.32pp 2.09 2.23pp 2.12 2.11 2.122. SDO 2.57 1.00 — .54pp .33pp .45pp .58pp 2.36pp 2.27pp 2.32pp 2.38pp 2.05 2.113. RWA–Total 2.98 0.80 — .83pp .86pp .86pp 2.45pp 2.24pp 2.41pp 2.37pp 2.11 2.19p

4. RWA–Traditionalism 2.80 0.88 — .63pp .52pp 2.35pp 2.09 2.37pp 2.24pp 2.09 2.22pp

5. RWA–Submission 2.74 0.86 — .60pp 2.39pp 2.25pp 2.41pp 2.33pp 2.12 2.136. RWA–Aggression 3.39 1.08 — 2.39pp 2.25pp 2.27pp 2.34pp 2.09 2.137. Item-based valence 1.58pp 1.40 — .44pp .46pp .39pp .17p .098. Item-based success 0.16 1.28 — .33pp .50pp .08 .23pp

9. General valence 3.47pp 2.23 — .55pp .17p .1010. General success 0.83pp 2.15 — .14 .21pp

11. Valence IAT 0.61 0.36 — .24pp

12. Success IAT 0.50 0.35 —

Note. N 5 146, 147, and 150 for the success Implicit Association Test (IAT), valence IAT, and explicit measures. Positive (negative) means indicate apositive evaluation of the democratic (autocratic) leadership concept. SDO 5 social dominance orientation; RWA 5 right-wing authoritarianism.p p , .05. pp p , .01.

7UNCERTAINTY AND LEADERSHIP PREFERENCES

pendent from each other but were interrelated (r 5 –.31).5 We

created a new variable of conjoint self-esteem by aggregating both

(z-standardized) variables, and we investigated the interactional

effect with (un)certainty on leadership evaluations. On the other

hand, a number of studies have demonstrated a significant Self-

Esteem Level 3 Self-Esteem Stability interaction with regard to a

variety of phenomena (for an overview, see Kernis, 2005). To do

justice to both perspectives, we tested a 2 (state @un#certainty) 3 2

(self-esteem level) 3 2 (self-esteem stability) model as well as our

predicted 2 (state @un#certainty) 3 2 (conjoint of self-esteem level

and stability) model.

We particularly expected differential uncertainty reactions to be

evident in changes of cognitive success expectations, but not in

changes of affective valence judgments, as these rely by definition

on feelings toward the attitude object and are relatively indepen-

dent of cognitive processes, that is, the evaluation of success. In

addition, as a preference for autocratic leadership is inconsistent

with the predominant social norm of democratic values in Western

societies, we expected stronger effects on implicit than on explicit

success measures.

Method

Participants and design. One hundred and twenty-eight stu-

dents from a German university participated in a 15-min study on

“Self-Perception” in return for 1 Euro and a chocolate bar. Up to

six participants at a time were present in the laboratory. One

participant failed to complete the uncertainty manipulation and

was excluded from further analyses, resulting in a total sample size

of 127 participants (78 women; Mage 5 22.0 years). On average,

participants stated their political orientation to be slightly more

liberal than conservative (1 5 extremely liberal, 9 5 extremely

conservative), M 5 4.36 (SD 5 1.29). As a first step, we tested a

2 (state @un#certainty: certain vs. uncertain) 3 2 (self-esteem level:

high vs. low) 3 2 (self-esteem stability: high vs. low) factorial

design. Because we expected additive effects of self-esteem level

and stability, in a second step we tested a 2 (state @un#certainty:

certain vs. uncertain) 3 2 (conjoint self-esteem: high level 1 high

stability vs. low level 1 low stability) factorial design. Participants

were randomly assigned to (un)certainty conditions, whereas self-

esteem level and stability were measured.

Procedure. After signing informed consent and providing

standard demographic information, participants were primed with

state certainty versus state uncertainty. Second, we controlled for

potential mood effects. Third, participants completed the implicit

leadership measures. To prevent carryover effects, this study dif-

fered from Studies 1a and 1b in that implicit always preceded

explicit leadership measures and that participants were randomly

assigned to one of the two IATs (valence IAT vs. success IAT).

Fourth, the explicit general leadership measures were administered

to all participants. Fifth, participants completed personality ques-

tionnaires regarding level and stability of self-esteem as well as

emotional uncertainty (EU), the latter to control for the influence

of trait uncertainty. Finally, participants filled in a manipulation

check on uncertainty.

Independent variables.(Un)certainty priming. We manipulated (un)certainty by ask-

ing participants to respond to the following two questions (adapted

from van den Bos, 2001): “Please briefly describe the emotions

that the thought of you being (un)certain of yourself arouses in

you,” and “Please write down what you think physically happens

to you as you feel (un)certain of yourself.”

Self-esteem level. Level of self-esteem was assessed by a

German 10-item version of the Rosenberg’s (1965) Self-Esteem

Inventory. A sample item (a 5 .90) was as follows: “I feel that I

am a person of worth, at least on an equal basis with others.”

Responses were provided on 4-point rating scales (1 5 strongly

disagree, 4 5 strongly agree).

Self-esteem stability. Self-esteem stability was assessed by a

German 5-item version of the Labile Self-Esteem Scale (Dykman,

1998; see also De Cremer & Sedikides, 2005). A sample item (a 5

.91) was as follows: “Compared to most people, my self-esteem

shifts rapidly from feeling good about myself one day to feeling

bad about myself the next day.” Five-point Likert scales were used

(1 5 strongly disagree, 5 5 strongly agree). Items were reverse

coded so that higher values represented higher stability instead of

lability.

Conjoint self-esteem. We calculated conjoint self-esteem by

averaging z-standardized self-esteem level and self-esteem stabil-

ity scores. The correlation between self-esteem level and stability

was r(127) 5 .55, p , .001.6

Dependent variables.General explicit leadership measures. General explicit va-

lence and success evaluations of democratic and autocratic lead-

ership were assessed with the four subscales introduced in Study

1a: general valence of democratic leadership (a 5 .83) and auto-

cratic leadership (a 5 .90) as well as general success of demo-

cratic leadership (a 5 .78) and autocratic leadership (a 5 .75).

Difference scores of relative preference were computed by sub-

tracting autocratic from democratic scores. Because of their lower

reliability scores in Study 1b, item-based evaluations were not

assessed.

Implicit leadership measures. Participants completed either

the valence IAT (a 5 .70) or the success IAT (a 5 .77) as

described in Study 1b. As we were more interested in individual

differences than in the overall mean scores of the IATs, we kept a

standardized block order to reduce irrelevant error variance.

Control variables.Manipulation check on (un)certainty. We asked participants

to indicate whether they had thought more strongly of certainty or

uncertainty (1 5 certainty, 7 5 uncertainty) during the completion

of the (un)certainty manipulation.

Mood. Participants completed the Positive and Negative Af-

fect Schedule (PANAS-X; Watson, Clark, & Tellegen, 1988).

Twenty items assessed how they felt at the time by means of

5-point rating scales (1 5 strongly disagree, 5 5 strongly agree).

5 Note that we speak of self-esteem stability, that is, the opposite of

self-esteem instability, and therefore we predict a positive, and not a

negative, relation with self-esteem level.6 Note that our correlation was higher than that found in the meta-

analysis of Okada (2010), r 5 –.31. This finding can be explained by the

type of measurement applied. Whereas we assessed self-esteem stability

via a questionnaire, most of the studies in the meta-analysis used a repeated

measurement technique. Okada has already pointed out the differences

between different ways of assessing self-esteem stability.

8 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

Half of the items measured Positive Affect (PA; a 5 .84), and the

other half assessed Negative Affect (NA; a 5 .88).

EU. EU was assessed by a German version of a subscale of

the Uncertainty Response Scale (Greco & Roger, 2001). The 15

items assess the extent to which participants experience uncer-

tainty as emotionally threatening (a 5 .88), for instance, “When

uncertain about what to do next, I tend to feel lost” (1 5 strongly

disagree, 7 5 strongly agree).7

Results

Control variables.Manipulation check on (un)certainty. The manipulation

check on uncertainty was subjected to a multiple regression anal-

ysis by including (un)certainty, self-esteem level, and self-esteem

stability as predictors in Step 1, the two-way interaction terms in

Step 2, and the three-way interaction term in Step 3.8 The pre-

dicted main effect of uncertainty was significant, b 5 .54,

t(122) 5 7.39, p , .001, showing that individuals in the uncer-

tainty condition reported more uncertainty (M 5 5.03, SD 5 1.86)

than individuals in the certainty condition (M 5 2.65, SD 5 1.73).

In addition, only a significant main effect of self-esteem level

occurred, b 5 –.22, t(122) 5 –2.48, p 5 .01 (all other ps . .22).

A multiple regression analysis with (un)certainty and conjoint

self-esteem as predictors in Step 1, and the two-way interaction

term in Step 2, revealed significant main effects for (un)certainty,

b 5 .54, t(123) 5 7.40, p , .001, and conjoint self-esteem, b 5

–.19, t(123) 5 –2.53, p 5 .01. The interaction was not significant

( p . .39).

Mood. We checked whether the uncertainty manipulation

affected PA and NA by conducting two independent multiple

regression analyses including (un)certainty, self-esteem level, and

self-esteem stability as predictors in Step 1, the two-way interac-

tion terms in Step 2, and the three-way interaction term in Step 3.

Regarding NA, only a significant main effect for self-esteem

level occurred, b 5 –.22, t(123) 5 –2.13, p 5 .04 (all other ps .

.20). Regarding PA, a significant main effect for (un)certainty,

b 5 –.19, t(123) 5 –2.25, p 5 .03, and marginally significant

main effects for self-esteem level and stability occurred, b 5 .17,

t(123) 5 1.68, p 5 .10, and b 5 .18, t(123) 5 1.75, p 5 .08,

respectively. The two-way interactions were not significant (all

ps . .50); however, the three-way interaction yielded significance,

b 5 .33, t(119) 5 2.53, p 5 .01.9

Two separate multiple regression analyses on NA and PA with

(un)certainty and conjoint self-esteem as predictors in Step 1, and

the two-way interaction term in Step 2, revealed for NA only a

significant main effect of conjoint self-esteem, b 5 –.28, t(124) 5

–3.30, p 5 .001 (all other ps . .23). Regarding PA, a significant

main effect for (un)certainty, b 5 –.19, t(124) 5 –2.26, p 5 .03,

and conjoint self-esteem, b 5 .30, t(124) 5 3.61, p , .001,

occurred. The interaction, however, was not significant ( p . .93).

Importantly, although unexpected mood effects emerged, control-

ling for PA and NA did not affect any of the central outcomes in

Study 2.

Dependent variables.Explicit leadership measures. First, we conducted two inde-

pendent multiple regressions on explicit valence and success eval-

uations including (un)certainty, self-esteem level, and self-esteem

stability as predictors in Step 1, the two-way interaction terms in

Step 2, and the three-way interaction term in Step 3. As can be

seen in the left part of Table 2, only the relation of self-esteem

level and explicit success evaluations was significant, indicating

that the higher the self-esteem, the more democratic the evalua-

tions were.

In our hypotheses, we predicted the effects of self-esteem level

and stability to be additive under conditions of uncertainty. We

therefore next conducted two multiple regression analyses with

(un)certainty, conjoint self-esteem, and their interaction as predic-

tors. For explicit valence and success judgments, however, no

main or interaction effects occurred (cf. Table 2).

Implicit leadership measures. First, two multiple regression

analyses were run with implicit leadership associations as depen-

dent variables and (un)certainty, self-esteem level, and self-esteem

stability as well as their interactions as predictors.10 As can be seen

in the right part of Table 2, only a significant Self-Esteem Level 3

Self-Esteem Stability interaction in regard of implicit success

evaluations occurred.11 Importantly, however, the three-way inter-

action with (un)certainty was nonsignificant.

To test our additive model, we conducted two multiple regres-

sion analyses with the valence and the success IATs as dependent

measures and (un)certainty, conjoint self-esteem, and their inter-

action as predictors (cf. Table 2). Regarding implicit valence, there

were no main effects and, as well, the two-way interaction was not

significant. More importantly, however, with regard to implicit

success, the predicted (Un)certainty 3 Conjoint Self-Esteem in-

teraction was significant. Simple slope analyses revealed that

under uncertainty, associations of democratic leadership with suc-

cess for stable high SEs (11 SD) became stronger, b 5 .35,

t(57) 5 2.01, p 5 .05, whereas for unstable low SEs (21 SD), they

substantially declined, b 5 –.38, t(57) 5 –2.23, p 5 .03. Simple

slopes are depicted in Figures 1A and 1B, for the valence IAT and

the success IAT, respectively.

7 We ran all analyses with EU as a covariate, however, as controlling for

EU did not affect any of the central outcomes, we dropped it from the

reported analyses.8 Note that throughout this article, and following Aiken and West

(1991), we centered continuous scores on their means prior to the regres-

sion analyses.9 Simple slope analyses revealed lower levels of PA under uncertainty

for high SEs with unstable self-esteem, b 5 –.50, t(119) 5 –2.31, p 5 .02,

and for low SEs with stable self-esteem, b 5 –.53, t(119) 5 –2.10, p 5 .04.

However, no difference between conditions was found for high SEs with

stable self-esteem ( p . .77) or for low SEs with unstable self-esteem ( p .

.29). Note that these results are similar to those reported by Kernis, Cornell,

Sun, Berry, and Harlow (1993), who found a three-way interaction of SE

level, SE stability, and (positive/negative) feedback.10 Different degrees of freedom for analogous analyses stem from the

fact that four participants had to be excluded because their error rates in

either the compatible or incompatible block were greater than 30%.11 Simple slope analyses revealed less democratic implicit success

evaluations by low SEs with unstable self-esteem than by low SEs with

stable self-esteem, b 5 .44, t(57) 5 2.65, p 5 .01. However, no

difference between high SEs with unstable versus stable self-esteem

occurred ( p . .15).

9UNCERTAINTY AND LEADERSHIP PREFERENCES

Discussion

Study 2 set out to extend our previous findings by experimen-

tally manipulating feelings of (un)certainty. Whereas we did not

expect differences between participants with stable high and un-

stable low SEs when primed with certainty, this difference should

become evident when primed with uncertainty. Specifically, we

tested a multiplicative against an additive model: (a) (Un)cer-

tainty 3 Self-Esteem Level 3 Self-Esteem Stability as opposed to

(b) (Un)certainty 3 (Self-Esteem Level 1 Self-Esteem Stability).

The three-way interaction in the first model was far from signifi-

cance regarding both explicit and implicit leadership evaluations,

contradicting the assumption of a multiplicative interplay between

self-esteem level and self-esteem stability under uncertainty. In

contrast, the additive model yielded the predicted (Un)certainty 3

(Self-Esteem Level 1 Self-Esteem Stability) interaction on im-

plicit success evaluations. In line with Hypothesis 2, priming

uncertainty for stable high SEs increased success IAT effects even

further (relative to the certainty condition), suggesting a demo-

cratic reaction. In line with Hypothesis 3, inducing the same

feelings of uncertainty in unstable low SEs resulted in a substantial

decrease of success IAT effects, indicating a submissive reaction.

Importantly, the democratic and submissive reactions in leader-

ship evaluation could only be observed for cognitive success

expectations of leadership styles, but not for affective valence, and

only if measured implicitly, but not explicitly. That the results

were specific to the success IAT rules out alternative explanations

in terms of demand effects (e.g., De Houwer, Beckers, & Moors,

2007), and it simultaneously indicates that observing the demo-

cratic and submissive reactions is difficult when targeting mere

affective reactions or when using explicit questionnaires.

Study 3

Study 2 provided initial empirical evidence for the hypothesized

democratic versus submissive reaction to uncertainty in stable high

SEs versus unstable low SEs. Study 3 aimed at replicating the

submissive reaction of unstable low SEs (Hypothesis 3) by making

trait uncertainty salient. Although individuals with low and unsta-

ble self-esteem should be dispositionally less certain of themselves

(e.g., Campbell & Lavallee, 1993; Kernis & Waschull, 1995), they

should not be chronically aware of their uncertainty because this

Table 2

Hierarchical Multiple Regression Results in Study 2

Model

Explicit measures Implicit measures (IATs)

Valence Success Valence Success

b t p b t p b t p b t p

Multiplicative(Un)certainty (U) .05 0.60 .55 .06 0.72 .48 2.08 20.58 .57 2.02 20.13 .90Self-esteem level (SEL) .21 1.94 .05 .23 2.22 .03 2.02 20.15 .88 .12 0.78 .44Self-esteem stability (SES) 2.07 20.64 .52 2.15 21.38 .17 .04 0.28 .78 .13 0.86 .39U 3 SEL 2.09 20.60 .55 2.03 20.16 .87 .10 0.43 .67 .23 1.13 .27U 3 SES 2.09 20.59 .55 2.14 21.01 .32 2.14 20.67 .51 .28 1.49 .14SEL 3 SES .07 0.75 .45 .17 1.82 .07 .12 0.84 .41 2.35 22.83 .01U 3 SEL 3 SES .01 0.06 .95 2.04 20.28 .78 2.11 20.53 .60 2.07 20.31 .76

AdditiveU .05 0.57 .57 .06 0.67 .50 2.07 20.57 .57 2.02 20.14 .89Conjoint self-esteem (CSE) .12 1.36 .18 .08 0.87 .39 .02 0.13 .89 .21 1.63 .11U 3 CSE 2.15 21.21 .23 2.16 21.24 .22 2.06 20.34 .73 .53 2.98 .004

Note. IATs 5 Implicit Association Tests.

(A)

-0,2

0

0,2

0,4

0,6

0,8

1

Certainty UncertaintyImp

lici

t Le

ad

ers

hip

Va

len

ce E

va

lua

�o

ns

Stable High SEs

Unstable Low SEs

(B)

-0,2

0

0,2

0,4

0,6

0,8

1

Certainty UncertaintyImp

lici

t Le

ad

ers

hip

Su

cce

ss E

va

lua

�o

ns

Stable High SEs

Unstable Low SEs

Figure 1. Individually z-standardized valence Implicit Association Test

(IAT) effects (A) and success IAT effects (B) as a function of state (un)cer-

tainty and conjoint self-esteem in Study 2. Stable High SEs 5 individuals with

stable and high self-esteem; Unstable Low SEs 5 individuals with unstable

and low self-esteem.

10 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

would be quite maladaptive, entailing the risk of depression (Ed-

wards & Weary, 1993). We therefore supposed that healthy un-

stable low SEs usually tend to suppress thoughts about their

uncertainty to prevent themselves from getting depressed. How-

ever, when trait uncertainty is made salient to them, they should be

highly motivated to reduce their feelings and cognitions of uncer-

tainty. We expected that simply assessing trait uncertainty on a

questionnaire should raise the awareness of uncertainty. Conse-

quently, unstable low SEs should strive for uncertainty reduction

and should exhibit a submissive reaction compared with unstable

low SEs whose trait uncertainty is not made salient. No such

effects were expected for stable high SEs, as they are assumed to

be certain of themselves by disposition (e.g., Campbell & La-

vallee, 1993; Kernis & Waschull, 1995). Working on a trait un-

certainty questionnaire should leave them, therefore, mostly unaf-

fected.

More specifically, as in Study 2, the expected submissive reac-

tion of unstable low SEs should be evident in changes in cognitive

success expectations. In contrast, we did not expect stable high and

unstable low SEs to differ on the valence measures of leadership.

As the submissive reaction is inconsistent with widely shared

democratic values, we again predicted stronger effects on implicit

than on explicit success measures.

Method

Participants and design. One hundred participants (59 women;

Mage 5 22.5 years), mainly German university students, completed

the 20-min online survey on “Attitudes and Personality Measures.”

Participants were recruited through an online pool. In return they

received the chance to win one of 10 book coupons (worth 10

Euros each). Participants were slightly more liberal than conser-

vative according to the mean political orientation score (1 5

extremely liberal, 9 5 extremely conservative), M 5 4.15 (SD 5

1.41). We again applied a multiplicative and an additive model.

First, we tested a 2 (trait uncertainty: nonsalient vs. salient) 3 2

(self-esteem level: high vs. low) 3 2 (self-esteem stability: high

vs. low) factorial design. In a second step, we tested a 2 (trait

uncertainty: nonsalient vs. salient) 3 2 (conjoint self-esteem: high

level 1 high stability vs. low level 1 low stability) factorial

design. Participants were randomly assigned to uncertainty sa-

lience conditions, while self-esteem level and stability were mea-

sured.

Procedure. In the trait uncertainty nonsalient condition (control

group), randomly chosen participants first completed the implicit and

explicit leadership measures, after which they completed the ques-

tionnaires on the moderator variables of self-esteem level and stabil-

ity. Finally, they completed the personality questionnaire on trait

uncertainty and answered two questions regarding their current

thoughts of uncertainty. In the trait uncertainty salient condition,

participants first filled in the questionnaires regarding self-esteem

level and stability. Then, just before they worked on implicit and

explicit leadership measures, they completed the uncertainty scale and

the questions regarding current thoughts of uncertainty to make trait

uncertainty salient. Overall, half of the participants encountered the

success IAT, whereas the other half completed the valence IAT.

Independent variables.Uncertainty salience. Participants completed a German 15-

item version of the EU subscale of the Uncertainty Response Scale

(a 5 .93) and the following two questions: “Have you just thought

about uncertainty?” (1 5 not at all, 9 5 definitely) and “How

intensely did you think about uncertainty?” (1 5 very moderately,

9 5 very strongly). Both items were highly interrelated (r 5 .85,

p , .001) and combined into a single index.

Self-esteem level. Absolute level of self-esteem was assessed

by the German 10-item version of Rosenberg’s Self-Esteem In-

ventory (a 5 .93).

Self-esteem stability. Self-esteem stability was assessed by

the reverse coded German 5-item version of the Labile Self-

Esteem Scale (a 5 .92).

Conjoint self-esteem. Conjoint self-esteem scores were cal-

culated by averaging z-standardized self-esteem level and stability

scores. The correlation between self-esteem level and stability was

r(100) 5 .53, p , .001.

Dependent variables.Explicit leadership measures. All participants indicated their

general valence and success evaluations of democratic and auto-

cratic leadership yielding four subscales: general valence of dem-

ocratic leadership (a 5 .87) and autocratic leadership (a 5 .82), as

well as general success of democratic leadership (a 5 .87) and

autocratic leadership (a 5 .85).

Implicit leadership measures. Participants either completed

the valence IAT (a 5 .73) or the success IAT (a 5 .75), as in

Study 2.

Results

Explicit leadership measures. First, we conducted two in-

dependent multiple regression analyses on explicit valence and

success evaluations including trait uncertainty salience, self-

esteem level, and self-esteem stability as predictors in Step 1, the

two-way interaction terms in Step 2, and the three-way interaction

term in Step 3. As can be seen in Table 3, no significant main or

interaction effects occurred.

For the additive model, trait uncertainty salience and conjoint

self-esteem were entered in Step 1, and their interaction was

entered in Step 2. However, only a significant main effect of

conjoint self-esteem on leadership valence evaluations emerged

(cf. Table 3).

Implicit leadership measures. First, two multiple regression

analyses were run with implicit leadership associations as depen-

dent variables and trait uncertainty salience, self-esteem level, and

self-esteem stability as well as their interactions as predictors.12 As

can be seen in Table 3, there were no significant main or interac-

tion effects in the multiplicative model.

To test our additive model, we conducted two multiple regres-

sion analyses with the valence IAT and the success IAT as depen-

dent measures. Trait uncertainty salience and conjoint self-esteem

were predictors in Step 1 and their interaction in Step 2. Results are

displayed in Table 3. With regard to implicit valence, there were

no main effects and, as well, the two-way interaction was not

significant. More importantly, however, with regard to implicit

success, the predicted Uncertainty Salience 3 Conjoint Self-

Esteem interaction was significant. Simple slope analyses revealed

12 Different degrees of freedom for analogous analyses stem from the

fact that five participants had to be excluded because their error rate in

either the compatible or incompatible block was greater than 30%.

11UNCERTAINTY AND LEADERSHIP PREFERENCES

that for stable high SEs (11 SD) there was no difference between

the salience and the no-salience condition ( p . .49); however, for

unstable low SEs (–1 SD), the association between success and

democratic leadership substantially declined when trait uncertainty

was made salient, b 5 –.63, t(42) 5 –3.17, p 5 .003. Simple

slopes are depicted in Figures 2A and 2B, for the valence IAT and

the success IAT, respectively.

Discussion

In Study 3, stable high SEs did not differ between conditions in

their evaluations on both explicit and implicit leadership measures,

as they had no reason to feel uncertain or react by changing their

preference toward autocratic leadership while working on an un-

certainty questionnaire. In contrast, unstable low SEs showed a

drop in their success IAT effects, relative to unstable low SEs in

the no salience condition, when we simply reminded them of their

trait uncertainty before completion of leadership measures. Over-

all, the findings are consistent with Hypothesis 3 and imply that

unstable low SEs are prone to a submissive reaction in response to

introspective processes.

Study 4

Up to this point, we focused on the circumstances that let

autocratic leadership appear successful in the eye of a follower.

Leadership, however, is always a two-sided interaction between

those who are led and those who lead. In Study 4, we investigated

whether the perspective of a follower versus that of a leader would

create a difference. We were particularly interested in how indi-

viduals in a leadership position react to uncertainty.

Leadership is the process of influencing others (Chemers, 2001).

The ability to influence others, in turn, is provided and sustained

by power (Galinsky, Gruenfeld, & Magee, 2003; Keltner, Gruen-

feld, & Anderson, 2003). Empirical evidence suggests that high

power increases confidence in one’s thoughts and perspectives (C.

Anderson & Galinsky, 2006; Brinol, Petty, Valle, Rucker, &

Becerra, 2007), heightens one’s orientation toward action (Galin-

sky et al., 2003), and makes one less vulnerable to situational and

interpersonal pressures (Galinsky, Magee, Gruenfeld, Whitson, &

Liljenquist, 2008). In other words, power both provides general

confidence and promotes a sense of being an effective actor.

Therefore, the expectation that they can successfully reduce un-

Table 3

Hierarchical Multiple Regression Results in Study 3

Model

Explicit measures Implicit measures (IATs)

Valence Success Valence Success

b t p b t p b t p b t p

MultiplicativeUncertainty salience (US) 2.05 20.47 .64 .09 0.85 .40 2.03 20.22 .83 2.26 21.77 .08Self-esteem level (SEL) .16 1.38 .17 .09 0.77 .44 .05 0.29 .78 .36 1.63 .11Self-esteem stability (SES) .14 1.20 .23 2.003 20.03 .98 .22 1.41 .17 2.10 20.47 .64US 3 SEL 2.18 20.91 .36 .04 0.20 .85 .22 0.82 .42 .22 0.79 .43US 3 SES .11 0.68 .50 2.12 20.72 .48 2.05 20.19 .85 .33 1.28 .21SEL 3 SES .07 0.59 .55 .008 0.07 .94 .07 0.37 .71 2.12 20.70 .49US 3 SEL 3 SES .14 0.90 .37 2.005 20.04 .97 .36 1.23 .23 2.05 20.24 .81

AdditiveUS 2.05 20.47 .64 .09 0.85 .40 2.03 20.22 .83 2.24 21.64 .11Conjoint self-esteem (CSE) .26 2.67 .01 .08 0.77 .45 .23 1.55 .13 .24 1.63 .11US 3 CSE 2.02 20.12 .91 2.06 20.44 .66 .13 0.56 .58 .46 2.69 .01

Note. IATs 5 Implicit Association Tests.

(A)

-0,2

0

0,2

0,4

0,6

0,8

1

Uncertainty Non-

Salient

Uncertainty SalientImp

lici

t Le

ad

ers

hip

Va

len

ce E

va

lua

�o

ns

Stable High SEs

Unstable Low SEs

(B)

-0,2

0

0,2

0,4

0,6

0,8

1

Uncertainty Non-

Salient

Uncertainty SalientImp

lici

t Le

ad

ers

hip

Su

cce

ss E

va

lua

�o

ns

Stable High SEs

Unstable Low SEs

Figure 2. Individually z-standardized valence Implicit Association Test

(IAT) effects (A) and success IAT effects (B) as a function of trait

uncertainty salience and conjoint self-esteem in Study 3. Stable High

SEs 5 individuals with stable and high self-esteem; Unstable Low SEs 5

individuals with unstable and low self-esteem.

12 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

certainty should be generally higher for individuals in a leadership

position compared with individuals in a follower position.

Does that mean that both stable high and unstable low SEs will

show a democratic reaction to uncertainty when in a leadership

position? It is possible, but previous research indicates that indi-

viduals with high power tend to rely on general knowledge struc-

tures such as preexisting views and values (Brinol et al., 2007).

They do not generally lack the capacity or motivation to process

information extensively but tend to distribute their attention ac-

cording to a sufficiency principle, that is, they only elaborate as

much as is necessary for goal attainment (Guinote, 2007a, 2007b;

Overbeck & Park, 2001; Vescio, Snyder, & Butz, 2003). The

amount of elaboration, in turn, is determined by level of confi-

dence. Consequently, the heightened level of confidence while in

a leadership position should render deliberative processes under

uncertainty irrelevant, so that both stable high and unstable low

SEs should rely on the dominant knowledge structures and the

most accessible information, such as preexisting values, that is,

their default attitude—a preference for democratic leadership.

Thus, we postulated that feelings of uncertainty lead to differential

changes in leadership evaluations for followers but not for leaders

(Hypothesis 4).

Method

Participants and design. One-hundred and sixty-seven stu-

dents from a German university participated in a 20-min study on

“Leadership” in return for 2.50 Euros and a chocolate bar. Up to 10

participants at a time were present in the laboratory. Seven par-

ticipants either failed to complete the IAT or explicit leadership

measures and were excluded from further analyses, resulting in a

total sample size of 160 participants (105 women; Mage 5 23.1

years). On average, participants had a slightly more liberal than

conservative political orientation (1 5 extremely liberal, 9 5

extremely conservative), M 5 4.18 (SD 5 1.29). As Studies 2 and

3 did not show any evidence for a multiplicative model, in Study

4 we focused on the predicted additive model, specifically we

tested a 2 (state @un#certainty: certain vs. uncertain) 3 2 (conjoint

self-esteem: high level 1 high stability vs. low level 1 low

stability) 3 2 (position: leader vs. follower) factorial design, with

participants randomly assigned to conditions, while self-esteem

level and stability were measured.

Procedure. First, participants completed the personality

questionnaires assessing self-esteem level and stability as well as

EU. Second, state uncertainty versus state certainty was experi-

mentally manipulated. Third, we administered PA and NA scales

to control for mood effects. Fourth, the leader versus follower

position was primed. Fifth, participants completed the implicit

followed by the explicit leadership measures. As Studies 2 and 3

already demonstrated that affective evaluations of leadership were

not subjected to influences of uncertainty, and the assessment of

implicit leadership associations is time consuming, we confined

implicit measures in Study 4 to the success IAT. The concepts of

RWA and SDO are related to the leader–follower distinction.

RWA expresses beliefs in traditional norms, obedience to author-

ities, and coercive social control. SDO represents beliefs in social

and economic inequality and the right of high status groups to

dominate weaker ones. Thus, RWA refers to submissiveness to-

ward authorities (e.g., leaders), whereas SDO refers to dominance

over others (e.g., followers). Sixth, we therefore assessed the RWA

and the SDO scale to control for their potential influence. Finally,

participants completed a manipulation check for the uncertainty

priming.

Independent variables.(Un)certainty priming. Half of the participants encountered

the uncertainty manipulation, whereas the other half were primed

with certainty. The procedure was the same as in Study 2, with one

exception of wording: “Please briefly describe the emotions that

the thought of one [instead of you] being (un)certain of oneself

[instead of you] arouses [instead of in you],” and “Please write

down what you think physically happens as one [instead of you]

feels (un)certain of oneself.”

Conjoint self-esteem. As before, self-esteem level was as-

sessed by the German 10-item version of the Rosenberg Self-

Esteem Inventory (a 5 .85). Self-esteem stability was assessed by

the reverse coded German five-item version of the Labile Self-

Esteem Scale (a 5 .92). Again, both scales were strongly inter-

related, r(160) 5 .52, p , .001. Conjoint self-esteem scores were

calculated by averaging z-standardized self-esteem level and sta-

bility scores.

Position priming. To manipulate the level of perceived power

experimentally, participants worked on one of the following tasks

just before IAT completion. Half of them were instructed as

follows: “During the completion of the following task [i.e., the

success IAT], please imagine another person with regard to whom

you are in a leadership position” (leader position). The other half

were told the following: “During the completion of the following

task [i.e., the success IAT], please imagine another person who is

in a leadership position with regard to you” (follower position).

Dependent variables.Explicit leadership measures. All participants indicated their

general explicit valence and success evaluations of democratic and

autocratic leadership, resulting in four reliable subscales (general

valence of democratic leadership: a 5 .87, and general valence of

autocratic leadership: a 5 .90; general success of democratic

leadership: a 5 .87, and general success of autocratic leadership:

a 5 .80).

Implicit leadership measure. All participants completed the

success IAT (a 5 .81) as described before.

Control variables.Manipulation check on (un)certainty. In Study 2, we as-

sessed the perceived uncertainty with a single item. To increase

reliability, participants in Study 4 were asked to remember and

indicate how they felt during the completion of the uncertainty task

on 10 items such as uncertain and torn, or certain and calm

(adapted from McGregor et al., 2001). Responses were given on

7-point rating scales (1 5 not at all, 7 5 very strongly). The

manipulation check on uncertainty was scaled such that higher

values indicated higher uncertainty (a 5 .91).

Mood. All participants completed a German version of the

short form of the PANAS-X (Mackinnon et al., 1999), with five

items measuring PA (a 5 .67) and five items measuring NA

(a 5 .78).

EU. Trait uncertainty was again assessed by the German

15-item version of the EU subscale of the Uncertainty Response

Scale (a 5 .91).

RWA and SDO. Participants completed a German 12-item

version of Altemeyer’s (1988) RWA scale on RWA (a 5 .77)

13UNCERTAINTY AND LEADERSHIP PREFERENCES

and the SDO scale on SDO (a 5 .85), the latter as described in

Study 1b.

Results

Control variables.Manipulation check on (un)certainty. The manipulation

check on uncertainty was subjected to a multiple regression anal-

ysis by including (un)certainty, conjoint self-esteem, and position

as predictors in Step 1, the two-way interaction terms in Step 2,

and the three-way interaction term in Step 3. The predicted main

effect of uncertainty was significant, b 5 .21, t(154) 5 2.73, p 5

.01, showing that individuals in the uncertainty condition reported

more uncertainty (M 5 3.00, SD 5 1.21) than individuals in the

certainty condition (M 5 2.56, SD 5 0.90). In addition, only a

significant main effect of conjoint self-esteem level occurred, b 5

–.30, t(154) 5 –3.96, p , .001. The main effect of position and the

interactions were not significant (all ps . .35).

Mood. We checked whether the uncertainty manipulation

affected PA and NA by conducting two independent multiple

regression analyses. Regarding PA, only a marginal significant

main effect for conjoint self-esteem level occurred, b 5 .14,

t(156) 5 1.82, p 5 .07. All other main and interaction effects were

not significant (all ps . .69).

Regarding NA, a marginally significant main effect of uncer-

tainty occurred, b 5 .14, t(156) 5 1.91, p 5 .06, and a significant

main effect of conjoint self-esteem occurred, b 5 –.36, t(156) 5

–4.88, p , .001. In Step 2 of the regression, the Conjoint Self-

Esteem 3 Position interaction yielded significance, b 5 .26,

t(153) 5 2.55, p 5 .01, and in Step 3, the three-way interaction

was marginally significant, b 5 .26, t(152) 5 1.66, p 5 .10 (all

other ps . .32). Importantly, controlling for the PA and the NA

scales, however, did not affect any other results.

EU, RWA and SDO. Controlling for EU, RWA, or SDO did

not affect any of the central outcomes and are therefore not

reported below.

Dependent variables.Explicit leadership measures. We analyzed the affective

and cognitive dimensions of leadership evaluation by including

(un)certainty, conjoint self-esteem, position, and their respec-

tive interactions in two separate multiple regression analyses.

However, for explicit valence and success judgments, no main

or interaction effects occurred (all ps . .18).

Implicit leadership measure. We applied the same regression

analysis to the success IAT effects.13 Main effects were absent (all

ps . .21), and the two-way interactions failed level of significance

(all ps . .12). Importantly, however, the (Un)certainty 3 Conjoint

Self-Esteem 3 Position interaction was significant, b 5 –.49,

t(145) 5 –2.81, p 5 .006.

As expected, in the follower’s position, for stable high SEs a

positive simple slope emerged, b 5 .33, t(145) 5 2.11, p 5 .04,

indicating a democratic reaction under uncertainty. Conversely, for

unstable low SEs a negative simple slope occurred, b 5 –.36,

t(145) 5 –2.22, p 5 .03, implying a submissive reaction. By

contrast, when participants had been primed with a leadership

position, simple slopes for both stable high SEs and unstable low

SEs were nonsignificant ( ps ..18). Thus, implicit success evalu-

ations did not change under uncertainty when a leadership position

was primed. Simple slopes are depicted in Figures 3A and 3B, for

the follower and the leader position, respectively.

Discussion

The results of Study 4 are in line with Hypotheses 2–4. Com-

pared with conditions of certainty, uncertainty paired with a fol-

lower (low power) position resulted in a democratic reaction for

stable high SEs (Hypothesis 2) and a submissive reaction for

unstable low SEs (Hypothesis 3). It is noteworthy that explicit

leadership measures again were not at all sensitive to these differ-

ential reactions. Importantly, the replicable two-way interaction

between uncertainty and conjoint self-esteem was cancelled out

when participants imagined themselves being in a leadership (high

power) position (Hypothesis 4). This result suggests a buffering

effect of high power against the consequences of uncertainty and

particularly against an enhanced preference for autocratic leader-

ship. For at least two reasons, this finding seems to be functional

with respect to the group’s goal. On the one hand, too much power

can make a leader corrupt (Keltner et al., 2003; Kipnis, 1972), and

an autocratic leadership style distributes the power in the group

very asymmetrically with the leader being in control. Therefore,

leaders who prefer an autocratic leadership style per se incur the

risk of power exploitation. On the other hand, changing one’s

leadership style whenever one feels uncertain also seems very

dysfunctional with regard to the group’s goal. Although contin-

gency models of leadership (F. E. Fiedler, 1964; Vroom & Yetton,

1973) argue that a leader should take the situation into account

when deciding how to lead, they refer to objective situational

factors and not to subjective situational feelings of the leader.

Therefore, a stable leadership style should be more efficient until

the objective situation calls for a change.

General Discussion

Viewed together, these findings strongly suggest that although

certainty produces a prodemocratic default attitude, uncertainty

triggers preferences for different kinds of leadership depending on

one’s self-esteem level and stability. To reiterate, in Studies 1a and

1b we developed and validated new explicit and implicit measures

on perceived valence and success of leadership. Under conditions

of certainty, participants strongly preferred democratic over auto-

cratic leadership with regard to both dimensions on implicit and

explicit leadership measures (Hypothesis 1). Employing the new

leadership measures in subsequent studies revealed a consistent

pattern of results. Uncertainty led to a democratic reaction for

stable high SEs (Hypothesis 2) and a submissive reaction for

unstable low SEs (Hypothesis 3). This replicable pattern is under-

scored by the application of different uncertainty manipulations:

priming state uncertainty as a situational factor (Study 2 and 4) and

making trait uncertainty salient as an interindividual difference

variable (Study 3). Taking into account that in all four studies

participants reported being slightly more liberal than conservative,

the submissive reaction of unstable low SEs is remarkable. Beyond

this, and in line with Hypothesis 4, the democratic versus submis-

13 Differences in degrees of freedom are due to the exclusion of seven

participants (error rates greater than 30% in the critical blocks).

14 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

sive reactions were erased when individuals imagined being in a

leadership position compared with a follower position. Impor-

tantly, the democratic and submissive reactions to uncertainty were

specific to the cognitively based success evaluations but not to the

affect-related, valence dimension of leadership and were demon-

strated only through implicit and not explicit measures.

Deliberation and Automaticity—A Paradox?

The democratic and submissive reactions to uncertainty for

stable high SEs and unstable low SEs were found only on implicit

measures that are supposed to reflect predominantly associative

processes. They should, therefore, at least to a certain degree, be

automatic and sensitive to situational variations. The theoretical

framework of the current article, however, was based on deliber-

ative outcome expectancy considerations, produced by a motive of

uncertainty reduction (Weary & Edwards, 1996). At first sight,

automaticity and deliberate reasoning seem to be contradictory,

especially if regarded as mutually exclusive and opposite ends of

a continuum (Shiffrin & Schneider, 1977). Though they are often

considered conceptually different processes, automatic and con-

trolled processing interact to bring about behavior, and this is true

for almost any implicit measure to date (K. Fiedler, Bluemke, &

Unkelbach, 2009). In our studies, participants were always at first

confronted with an uncertainty manipulation, and they thought

about uncertainty for at least a few minutes. Drawing on Weary

and Edwards’s (1996) causal uncertainty model, during this time a

“comparator” identifies the discrepancy between the current state

of uncertainty and the desired state of certainty. It activates po-

tential action plans to reduce this discrepancy. Action plans are

then analyzed by the “outcome expectancy assessor.” The crucial

point that differentiates stable high SEs from unstable low SEs is

whether they believe that they can efficiently reduce uncertainty

themselves. Whereas stable high SEs are prepared to act, unstable

low SEs doubt their abilities. Confronted with two leadership

styles that offer different degrees of participation, differential

preactivation results in different outcome expectancies. Thus, the

outcomes of the implicit measure may show characteristics of

automaticity, but they reflect the consequences of a deliberative

process of weighing expectancies that changes the underlying

associations in memory (cf. Gawronski & Bodenhausen, 2006).

Why Is the Difference Between Implicit and ExplicitImportant?

We found across studies that uncertainty led stable high SEs to

more democratic and unstable low SEs to more autocratic implicit

evaluations of success compared with conditions of certainty.

Explicit evaluations of success, however, remained largely unaf-

fected. In their associative-propositional evaluation model,

Gawronski and Bodenhausen (2006) argued that associative (im-

plicit) evaluations are examined with respect to their consistency

with activated propositions and are rejected as invalid for explicit

judgments in the case of inconsistencies. Following this reasoning,

the associations between autocratic 1 success and democratic 1

failure conflict with propositions belonging to a general belief

system of democratic values. The submissive reaction to uncer-

tainty for unstable low SEs was found consistently only on implicit

measures assessing the automatic evaluations of leadership suc-

cess. In contrast, the associations between democratic 1 success

and autocratic 1 failure are in line with predominant democratic

values. Increased democratic preferences should therefore be re-

garded as valid and the democratic reaction of stable high SEs

should be expressed on implicit and explicit measures. However,

the democratic reaction was also found only on implicit and not on

explicit leadership measures.

To explain this result, we draw on Strack and Deutsch’s (2004)

reflective–impulsive model. The authors have distinguished be-

tween impulsive and reflective processes. The impulsive system is

conceived as a simple associative network (see also, J. R. Ander-

son & Bower, 1973) in which spreading activation is induced by

perceptual input or reflective processes. The activation of an

element spreads to other elements in proportion to the strength of

the links between them. Activation varies and information is only

retrieved for further elaboration if the activation level exceeds a

certain threshold. Thus, whether an element enters the reflective

system depends on its accessibility, that is, its excitation level.

Without frequent and recent activation, however, the activation

level decays (cf. Higgins, 1996) and in consequence the informa-

tion becomes less accessible for the reflective system.

In our studies, the democratic and submissive reactions may not

have reached the threshold for further processing and were, there-

fore, only exhibited on implicit but not on explicit measures. This

(A)

-0,2

0

0,2

0,4

0,6

0,8

1

Certainty UncertaintyImp

lici

t L

ea

de

rsh

ip S

ucc

ess

Ev

alu

a�

on

s

Stable High SEs

Unstable Low SEs

(B)

-0,2

0

0,2

0,4

0,6

0,8

1

Certainty UncertaintyImp

lici

t Le

ad

ers

hip

Su

cce

ss E

va

lua

�o

ns

Stable High SEs

Unstable Low SEs

Figure 3. Individually z-standardized success Implicit Association Test

(IAT) effects as a function of uncertainty and conjoint self-esteem for

participants in a follower position (A) and a leadership position (B) in

Study 4. Stable High SEs 5 individuals with stable and high self-esteem;

Unstable Low SEs 5 individuals with unstable and low self-esteem.

15UNCERTAINTY AND LEADERSHIP PREFERENCES

can be explained by the circumstance that awareness of uncertainty

is an exceptional state that occurs rarely in everyday life because

individuals avoid thoughts of uncertainty if possible. However,

more enduring and stronger states of uncertainty might heighten

the accessibility of matching associative patterns, that is, the

democratic reaction for stable high SEs and the submissive reac-

tion for unstable low SEs.

Future Research

Although our studies provide converging evidence in support of the

main hypotheses, some questions remain that call for further investi-

gation. First, these studies relied on two different methods of inducing

uncertainty: priming state uncertainty and making trait uncertainty

salient. However, it seems important to put the hypotheses to the test

with even broader manipulations of uncertainty such as, for example,

mortality salience. In line with terror management theory (e.g., Cohen

et al., 2005, 2004; Landau et al., 2004), even stronger effects on

leadership evaluations for more threatening stimuli might be postu-

lated. The democratic and submissive reactions found only on implicit

measures in the current studies might then enter the reflective system

and be expressed also on explicit measures. On the other hand, our

findings show that even mild cues of uncertainty can have important

effects that become visible only with the help of association-based

measurement procedures.

Second, in Study 4 we showed that assuming a leadership

perspective was an effective buffer against the reactions to uncer-

tainty. We pointed out that this is an important finding because a

leader who prefers an autocratic leadership style whenever he/she

feels uncertain might incur the risk of power exploitation. In a

recent study, Maner and Mead (2010) found that the combination

of high dominance motivation and instability of the group’s hier-

archy led to self-interested and group-damaging actions by the

leader. Though SDO and RWA did not affect our results in Study

4, it seems worthwhile to investigate the influence of dominance

motivation on a leader’s reaction to uncertainty.

Third, because of its comparative nature, the traditional IAT is

ambiguous in absolute evaluation of the respective target concepts

(Blanton, Jaccard, Gonzales, & Christie, 2006; K. Fiedler, Messner, &

Bluemke, 2006; Nosek et al., 2005). Therefore, it is difficult to infer

from the current findings whether the democratic and submissive

reactions are caused by changes in the evaluation of democratic or of

autocratic leadership or of both at the same time. Wigboldus, Holland,

and van Knippenberg (2004) introduced the Single-Target IAT, which

presents participants with only one of the target categories, while both

attribute categories remain. Given the reliability and validity of such

single-category procedures (e.g., Bluemke & Friese, 2008), it seems

to be a promising tool for disentangling the intricate pattern of the

democratic and submissive reactions.

Finally, although the relationship between attitudes and behav-

ior is complex and sometimes difficult to uncover, it seems worth-

while to investigate whether the democratic and submissive reac-

tions demonstrated on implicit measures can influence actual

behavior such as voting for political candidates, support and do-

nations for democratic institutions, or choice of an employer.

Concluding Comments

The goal of the current research was to determine the circum-

stances under which democratic leadership is judged as less suc-

cessful even if more pleasant, while autocratic leadership is judged

as more successful even if unpleasant. Viewed together, these

studies considerably extend the threat-uncertainty, self-esteem,

and leadership literatures by showing that uncertainty, self-esteem

level and stability, as well as leader versus follower perspective

have a moderating impact on individuals’ leadership evaluations.

References

Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N.

(1950). The authoritarian personality. New York, NY: Harper & Row.

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and

interpreting interactions. New York, NY: Sage.

Allport, G. W. (1954). The nature of prejudice. Reading, MA: Addison-

Wesley.

Altemeyer, B. (1988). Enemies of freedom: Understanding right-wing

authoritarianism. San Francisco, CA: Jossey-Bass.

Altemeyer, B. (1996). The authoritarian specter. Cambridge, MA: Harvard

University Press.

Anderson, C., & Galinsky, A. D. (2006). Power, optimism, and risk-taking.

European Journal of Social Psychology, 36, 511–536. doi:10.1002/ejsp.324

Anderson, J. R., & Bower, G. H. (1973). Human associative memory.

Washington, DC: Winston & Sons.

Aspinwall, L. G., & Taylor, S. E. (1992). Modeling cognitive adaptation:

A longitudinal investigation of the impact of individual differences and

coping on college adjustment and performance. Journal of Personality

and Social Psychology, 63, 989–1003. doi:10.1037/0022-3514.63.6.989

Bass, B. M. (1990). Bass and Stogdill’s handbook of leadership: Theory,

research, and managerial applications (3rd ed.). New York, NY: Free

Press.

Baumeister, R. F. (1993). Self-esteem: The puzzle of low self-regard. New

York, NY: Plenum Press.

Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. D. (2003).

Does high self-esteem cause better performance, interpersonal success,

happiness, or healthier lifestyles? Psychological Science in the Public

Interest, 4, 1–44. doi:10.1111/1529-1006.01431

Blanton, H., Jaccard, J., Gonzales, P. M., & Christie, C. (2006). Decoding

the Implicit Association Test: Implications for criterion prediction. Jour-

nal of Experimental Social Psychology, 42, 192–212. doi:10.1016/

j.jesp.2005.07.003

Bluemke, M., & Friese, M. (2006). Do features of stimuli influence IAT

effects? Journal of Experimental Social Psychology, 42, 163–176. doi:

10.1016/j.jesp.2005.03.004

Bluemke, M., & Friese, M. (2008). Reliability and validity of the Single-

Target IAT (ST-IAT): Assessing automatic affect towards multiple

attitude objects. European Journal of Social Psychology, 38, 977–997.

doi:10.1002/ejsp.487

Brinol, P., Petty, R. E., Valle, C., Rucker, D. D., & Becerra, A. (2007). The

effects of message recipients’ power before and after persuasion: A

self-validation analysis. Journal of Personality and Social Psychology,

93, 1040–1053. doi:10.1037/0022-3514.93.6.1040

Brockner, J. (1983). Low self-esteem and behavioral plasticity: Some

implications. In L. Wheeler & P. R. Shaver (Eds.), Review of personality

and social psychology (4th ed., pp. 237–271). Beverly Hills, CA: Sage.

Brockner, J., Heuer, L., Siegel, P. A., Wiesenfeld, B. M., Martin, C.,

Grover, S., . . . Bjorgvinsson, S. (1998). The moderating effect of self-

esteem in reaction to voice: Converging evidence from five studies.

Journal of Personality and Social Psychology, 75, 394–407. doi:

10.1037/0022-3514.75.2.394

Brown, J. D. (1986). Evaluations of self and others: Self-enhancement

biases in social judgments. Social Cognition, 4, 353–376. doi:10.1521/

soco.1986.4.4.353

Brown, J. D., Collins, R. L., & Schmidt, G. W. (1988). Self-esteem and

16 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

direct versus indirect forms of self-enhancement. Journal of Personality

and Social Psychology, 55, 445–453. doi:10.1037/0022-3514.55.3.445

Campbell, J. D., & Lavallee, L. F. (1993). Who am I? The role of

self-concept confusion in understanding the behavior of people with low

self-esteem. In R. F. Baumeister (Ed.), Self-esteem: The puzzle of low

self-regard (pp. 3–20). New York, NY: Plenum Press.

Chemers, M. M. (2001). Leadership effectiveness: An integrative review.

In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social

psychology: Group processes (pp. 376–399). Oxford, England: Black-

well. doi:10.1002/9780470998458.ch16

Cohen, F., Ogilvie, D. M., Solomon, S., Greenberg, J., & Pyszczynski, T.

(2005). American roulette: The effect of reminders of death on support

for George W. Bush in the 2004 presidential election. Analyses of Social

Issues and Public Policy (ASAP), 5, 177–187. doi:10.1111/j.1530-

2415.2005.00063.x

Cohen, F., Solomon, S., Maxfield, M., Pyszczynski, T., & Greenberg, J.

(2004). Fatal attraction: The effects of mortality salience on evaluations

of charismatic, task-oriented, and relationship-oriented leaders. Psycho-

logical Science, 15, 846–851. doi:10.1111/j.0956-7976.2004.00765.x

Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-

determination in human behavior. New York, NY: Plenum Press.

De Cremer, D. (2007). Emotional effects of distributive justice as a

function of autocratic leader behavior. Journal of Applied Social Psy-

chology, 37, 1385–1404. doi:10.1111/j.1559-1816.2007.00217.x

De Cremer, D., & Sedikides, C. (2005). Self-uncertainty and responsive-

ness to procedural justice. Journal of Experimental Social Psychology,

41, 157–173. doi:10.1016/j.jesp.2004.06.010

Degner, J., Wentura, D., & Rothermund, K. (2006). Indirect assessment of

attitudes with response-time-based measures. Zeitschrift fur Sozialpsy-

chologie, 37, 131–139. doi:10.1024/0044-3514.37.3.131

De Houwer, J., Beckers, T., & Moors, A. (2007). Novel attitudes can be

faked on the Implicit Association Test. Journal of Experimental Social

Psychology, 43, 972–978. doi:10.1016/j.jesp.2006.10.007

Doty, R. M., Peterson, B. E., & Winter, D. G. (1991). Threat and author-

itarianism in the United States, 1978–1987. Journal of Personality and

Social Psychology, 61, 629–640. doi:10.1037/0022-3514.61.4.629

Duckitt, J., Bizumic, B., Krauss, S. W., & Heled, E. (2010). A tripartite

approach to right-wing authoritarianism: The authoritarianism-

conservatism-traditionalism model. Political Psychology, 31, 685–715.

doi:10.1111/j.1467-9221.2010.00781.x

Duckitt, J., & Fisher, K. (2003). The impact of social threat on world view

and ideological attitudes. Political Psychology, 24, 199 –222. doi:

10.1111/0162-895X.00322

Dykman, B. M. (1998). Integrating cognitive and motivational factors in

depression: Initial tests of a goal-orientation approach. Journal of Per-

sonality and Social Psychology, 74, 139 –158. doi:10.1037/0022-

3514.74.1.139

Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort

Worth, TX: Harcourt Brace Jovanovich.

Edwards, J. A., & Weary, G. (1993). Depression and the impression-

formation continuum: Piecemeal processing despite the availability of

category information. Journal of Personality and Social Psychology, 64,

636–645. doi:10.1037/0022-3514.64.4.636

Feldman, S. (2003). Enforcing social conformity: A theory of authoritari-

anism. Political Psychology, 24, 41–74. doi:10.1111/0162-895X.00316

Feldman, S., & Stenner, K. (1997). Perceived threat and authoritarianism.

Political Psychology, 18, 741–770. doi:10.1111/0162-895X.00077

Fiedler, F. E. (1964). A contingency theory of leadership effectiveness. In

L. Berkowitz (Ed.), Advances in experimental social psychology (pp.

149 –190). New York, NY: Academic Press. doi:10.1016/S0065-

2601(08)60051-9

Fiedler, F. E. (1967). Style of leadership and performance of coacting

groups. Zeitschrift fur Experimentelle und Angewandte Psychologie, 14,

200–217.

Fiedler, K., Bluemke, M., & Unkelbach, C. (2009). Exerting control over

allegedly automatic associative processes. In D. M. Tice (Ed.), Psychol-

ogy of self-regulation: Cognitive, affective, and motivational processes

(pp. 249–269). New York, NY: Psychology Press.

Fiedler, K., Messner, C., & Bluemke, M. (2006). Unresolved problems

with the “I,” the “A,” and the “T”: A logical and psychometric critique

of the Implicit Association Test (IAT). European Review of Social

Psychology, 17, 74–147. doi:10.1080/10463280600681248

Fleishman, E. A. (1973). Twenty years of consideration and structure. In E. A.

Fleishman & J. G. Hunt (Eds.), Current developments in the study of

leadership (pp. 1–37). Carbondale, IL: Southern Illinois University Press.

Galinsky, A. D., Gruenfeld, D. H., & Magee, J. C. (2003). From power to

action. Journal of Personality and Social Psychology, 85, 453–466.

doi:10.1037/0022-3514.85.3.453

Galinsky, A. D., Magee, J. C., Gruenfeld, D. H., Whitson, J. A., &

Liljenquist, K. A. (2008). Power reduces the press of the situation:

Implications for creativity, conformity, and dissonance. Journal of Per-

sonality and Social Psychology, 95, 1450–1466. doi:10.1037/a0012633

Gawronski, B., & Bodenhausen, G. V. (2006). Associative and proposi-

tional processes in evaluation: An integrative review of implicit and

explicit attitude change. Psychological Bulletin, 132, 692–731. doi:

10.1037/0033-2909.132.5.692

Gillath, O., & Hart, J. (2010). The effects of psychological security and

insecurity on political attitudes and leadership preferences. European

Journal of Social Psychology, 40, 122–134. doi:10.1002/ejsp.614

Greco, V., & Roger, D. (2001). Coping with uncertainty: The construction

and validation of a new measure. Personality and Individual Differ-

ences, 31, 519–534. doi:10.1016/S0191-8869(00)00156-2

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring

individual differences in implicit cognition: The Implicit Association

Test. Journal of Personality and Social Psychology, 74, 1464–1480.

doi:10.1037/0022-3514.74.6.1464

Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding

and using the Implicit Association Test: I. An improved scoring algo-

rithm. Journal of Personality and Social Psychology, 85, 197–216.

doi:10.1037/0022-3514.85.2.197

Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R.

(2009). Understanding and using the Implicit Association Test: III.

Meta-analysis of predictive validity. Journal of Personality and Social

Psychology, 97, 17–41. doi:10.1037/a0015575

Guinote, A. (2007a). Power affects basic cognition: Increased attentional

inhibition and flexibility. Journal of Experimental Social Psychology,

43, 685–697. doi:10.1016/j.jesp.2006.06.008

Guinote, A. (2007b). Power and goal pursuit. Personality and Social

Psychology Bulletin, 33, 1076–1087. doi:10.1177/0146167207301011

Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability,

and salience. In E. T. Higgins & A. W. Kruglanski (Eds.), Social

psychology: Handbook of basic principles (pp. 133–168). New York,

NY: Guilford Press.

Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M.

(2005). A meta-analysis on the correlation between the Implicit Asso-

ciation Test and explicit self-report measures. Personality and Social

Psychology Bulletin, 31, 1369–1385. doi:10.1177/0146167205275613

Hogg, M. A. (2000). Subjective uncertainty reduction through self-

categorization: A motivational theory of social identity processes. Eu-

ropean Review of Social Psychology, 11, 223–255. doi:10.1080/

14792772043000040

Hogg, M. A. (2007). Social psychology of leadership. In A. W. Kruglanski

& E. T. Higgins (Eds.), Social psychology: Handbook of basic principles

(2nd ed., pp. 716–733). New York, NY: Guilford Press.

Hogg, M. A. (2009). Managing self-uncertainty through group identifica-

tion. Psychological Inquiry, 20, 221–224. doi:10.1080/

10478400903333452

Jost, J. T., Napier, J. L., Thorisdottir, H., Gosling, S. D., Palfai, T. P., &

17UNCERTAINTY AND LEADERSHIP PREFERENCES

Ostafin, B. (2007). Are needs to manage uncertainty and threat associ-

ated with political conservatism or ideological extremity? Personality

and Social Psychology Bulletin, 33, 989 –1007. doi:10.1177/

0146167207301028

Judge, T. A., Piccolo, R. F., & Ilies, R. (2004). The forgotten ones? The

validity of consideration and initiating structure in leadership research.

Journal of Applied Psychology, 89, 36 –51. doi:10.1037/0021-

9010.89.1.36

Kagan, J. (1972). Motives and development. Journal of Personality and

Social Psychology, 22, 51–66. doi:10.1037/h0032356

Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach,

and inhibition. Psychological Review, 110, 265–284. doi:10.1037/0033-

295X.110.2.265

Kernis, M. H. (2005). Measuring self-esteem in context: The importance of

stability of self-esteem in psychological functioning. Journal of Person-

ality, 73, 1569–1605. doi:10.1111/j.1467-6494.2005.00359.x

Kernis, M. H., Cornell, D. P., Sun, C., Berry, A., & Harlow, T. (1993).

There’s more to self-esteem than whether it is high or low: The impor-

tance of stability of self-esteem. Journal of Personality and Social

Psychology, 65, 1190–1204.

Kernis, M. H., & Lakey, C. E. (2010). Fragile versus secure high self-

esteem: Implications for defensiveness and insecurity. In R. M. Arkin,

K. C. Oleson & P. J. Carroll (Eds.), Handbook of the uncertain self (pp.

360–378). New York, NY: Psychology Press.

Kernis, M. H., Paradise, A. W., Whitaker, D. J., Wheatman, S. R., &

Goldman, B. N. (2000). Master of one’s psychological domain? Not

likely if one’s self-esteem is unstable. Personality and Social Psychol-

ogy Bulletin, 26, 1297–1305. doi:10.1177/0146167200262010

Kernis, M. H., & Waschull, S. B. (1995). The interactive roles of stability

and level of self-esteem: Research and theory. In M. P. Zanna (Ed.),

Advances in experimental social psychology (Vol. 27, pp. 93–141). San

Diego, CA: Academic Press.

Kipnis, D. (1972). Does power corrupt? Journal of Personality and Social

Psychology, 24, 33–41. doi:10.1037/h0033390

Kosloff, S., Greenberg, J., Weise, D., & Solomon, S. (2010). Mortality

salience and political preferences: The roles of charisma and political

orientation. Journal of Experimental Social Psychology, 46, 139–145.

doi:10.1016/j.jesp.2009.09.002

Landau, M. J., Solomon, S., Greenberg, J., Cohen, F., Pyszczynski, T.,

Arndt, J., . . . Cook, A. (2004). Deliver us from evil: The effects of

mortality salience and reminders of 9/11 on support for President George

W. Bush. Personality and Social Psychology Bulletin, 30, 1136–1150.

doi:10.1177/0146167204267988

Lavine, H., Burgess, D., Snyder, M., Transue, J., Sullivan, J. L., Haney, B.,

& Wagner, S. H. (1999). Threat, authoritarianism, and voting: An

investigation of personality and persuasion. Personality and Social Psy-

chology Bulletin, 25, 337–347. doi:10.1177/0146167299025003006

LePine, J. A., & Van Dyne, L. (1998). Predicting voice behavior in work

groups. Journal of Applied Psychology, 83, 853–868. doi:10.1037/0021-

9010.83.6.853

Lewin, K., Lippitt, R., & White, R. K. (1939). Patterns of aggressive

behavior in experimentally created “social climates.” The Journal of

Social Psychology, 10, 269–299. doi:10.1080/00224545.1939.9713366

Likert, R. (1961). New patterns of management. New York, NY: McGraw-

Hill.

Lippitt, R. (1940). An experimental study of the effect of democratic and

authoritarian group atmospheres. University of Iowa Studies in Child

Welfare, 16, 43–195.

Loseman, A., Miedema, J., van den Bos, K., & Vermunt, R. (2009).

Exploring how people respond to conflicts between self-interest and

fairness: Influence of threats to the self on affective reactions to advan-

tageous inequity. Australian Journal of Psychology, 61, 13–21. doi:

10.1080/00049530802607605

Mackinnon, A., Jorm, A. F., Christensen, H., Korten, A. E., Jacomb, P. A.,

& Rodgers, B. (1999). A short form of the Positive and Negative Affect

Schedule: Evaluation of factorial validity and invariance across demo-

graphic variables in a community sample. Personality and Individual

Differences, 27, 405–416. doi:10.1016/S0191-8869(98)00251-7

Maner, J. K., & Mead, N. L. (2010). The essential tension between

leadership and power: When leaders sacrifice group goals for the sake of

self-interest. Journal of Personality and Social Psychology, 99, 482–

497. doi:10.1037/a0018559

Marsh, H. W. (1986). Self-serving effect (bias?) in academic attributions:

Its relation to academic achievement and self-concept. Journal of Edu-

cational Psychology, 78, 190–200. doi:10.1037/0022-0663.78.3.190

Maslow, A. H. (1954). The instinctoid nature of basic needs. Journal of

Personality, 22, 326–347. doi:10.1111/j.1467-6494.1954.tb01136.x

McCann, S. J. H. (1997). Threatening times, “strong” presidential popular

vote winners, and the victory margin, 1824–1964. Journal of Person-

ality and Social Psychology, 73, 160 –170. doi:10.1037/0022-

3514.73.1.160

McCann, S. J. H. (2008). Societal threat, authoritarianism, conservatism,

and U.S. state death penalty sentencing (1977–2004). Journal of Per-

sonality and Social Psychology, 94, 913–923. doi:10.1037/0022-

3514.94.5.913

McFarlin, D. B., & Blascovich, J. (1981). Effects of self-esteem and

performance feedback on future affective preferences and cognitive

expectations. Journal of Personality and Social Psychology, 40, 521–

531. doi:10.1037/0022-3514.40.3.521

McGregor, I., Zanna, M. P., Holmes, J. G., & Spencer, S. J. (2001).

Compensatory conviction in the face of personal uncertainty: Going to

extremes and being oneself. Journal of Personality and Social Psychol-

ogy, 80, 472–488. doi:10.1037/0022-3514.80.3.472

Miller, K. I., & Monge, P. R. (1986). Participation, satisfaction, and

productivity: A meta-analytic review. Academy of Management Journal,

29, 727–753. doi:10.2307/255942

Nail, P. R., & McGregor, I. (2009). Conservative shift among liberals and

conservatives following 9/11/01. Social Justice Research, 22, 231–240.

doi:10.1007/s11211-009-0098-z

Neuberger, O. (1972). Experimentelle Untersuchungen von Fuhrungsstilen

[Experimental study of leaderships]. Gruppendynamik, 3, 192–219.

Nielsen, M. E., & Miller, C. E. (1997). The transmission of norms regard-

ing group decision rules. Personality and Social Psychology Bulletin, 23,

516–525. doi:10.1177/0146167297235007

Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2005). Understanding

and using the Implicit Association Test: II. Method variables and con-

struct validity. Personality and Social Psychology Bulletin, 31, 166–

180. doi:10.1177/0146167204271418

Okada, R. (2010). A meta-analytic review of the relation between self-

esteem level and self-esteem instability. Personality and Individual

Differences, 48, 243–246.

Overbeck, J. R., & Park, B. (2001). When power does not corrupt: Superior

individuation processes among powerful perceivers. Journal of Person-

ality and Social Psychology, 81, 549 –565. doi:10.1037/0022-

3514.81.4.549

Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social

dominance orientation: A personality variable predicting social and

political attitudes. Journal of Personality and Social Psychology, 67,

741–763. doi:10.1037/0022-3514.67.4.741

Rickert, E. J. (1998). Authoritarianism and economic threat: Implications

for political behavior. Political Psychology, 19, 707–720. doi:10.1111/

0162-895X.00128

Rokeach, M. (1954). The nature and meaning of dogmatism. Psychological

Review, 61, 194–204. doi:10.1037/h0060752

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton,

NJ: Princeton University Press.

Rosenberg, M. (1979). Conceiving the self. New York, NY: Basic Books.

Ross, M., & Wilson, A. E. (2002). It feels like yesterday: Self-esteem,

18 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG

valence of personal past experiences, and judgments of subjective dis-

tance. Journal of Personality and Social Psychology, 82, 792–803.

doi:10.1037/0022-3514.82.5.792

Rutte, C. G., & Wilke, H. A. M. (1985). Preference for decision structures

in a social dilemma situation. European Journal of Social Psychology,

15, 367–370. doi:10.1002/ejsp.2420150311

Sales, S. M. (1972). Economic threat as a determinant of conversion rates

in authoritarian and nonauthoritarian churches. Journal of Personality

and Social Psychology, 23, 420–428. doi:10.1037/h0033157

Sales, S. M. (1973). Threat as a factor in authoritarianism: An analysis of

archival data. Journal of Personality and Social Psychology, 28, 44–57.

doi:10.1037/h0035588

Sales, S. M., & Friend, K. E. (1973). Success and failure as determinants

of level of authoritarianism. Behavioral Science, 18, 163–172. doi:

10.1002/bs.3830180304

Samuelson, C. D., & Messick, D. M. (1986). Inequities in access to and use

of shared resources in social dilemmas. Journal of Personality and

Social Psychology, 51, 960–967. doi:10.1037/0022-3514.51.5.960

Scheier, M. F., Carver, C. S., & Bridges, M. W. (1994). Distinguishing

optimism from neuroticism (and trait anxiety, self-mastery, and self-

esteem): A reevaluation of the Life Orientation Test. Journal of Person-

ality and Social Psychology, 67, 1063–1078. doi:10.1037/0022-

3514.67.6.1063

Sedikides, C., De Cremer, D., Hart, C., & Brebels, L. G. G. (2010).

Procedural fairness responses in the context of self-uncertainty. In R. M.

Arkin, K. C. Oleson, & P. J. Caroll (Eds.), Handbook of the uncertain

self (pp. 142–159). New York, NY: Psychology Press.

Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human

information processing: II. Perceptual learning, automatic attending and

a general theory. Psychological Review, 84, 127–190. doi:10.1037/0033-

295X.84.2.127

Sorrentino, R. M., Short, J. C., & Raynor, J. O. (1984). Uncertainty

orientation: Implications for affective and cognitive views of achieve-

ment behavior. Journal of Personality and Social Psychology, 46, 189–

206. doi:10.1037/0022-3514.46.1.189

Steffens, M. C., & Plewe, I. (2001). Items’ cross-category associations as

a confounding factor in the Implicit Association Test. Zeitschrift fur

Experimentelle Psychologie, 48, 123–134. doi:10.1026//0949-

3946.48.2.123

Stenner, K. (2005). The authoritarian dynamic. New York, NY: Cam-

bridge University Press. doi:10.1017/CBO9780511614712

Stogdill, R. M. (1974). Handbook of leadership: A survey of theory and

research. New York, NY: Free Press.

Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of

social behavior. Personality and Social Psychology Review, 8, 220–247.

doi:10.1207/s15327957pspr0803_1

Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social

psychological perspective on mental health. Psychological Bulletin, 103,

193–210. doi:10.1037/0033-2909.103.2.193

Tyler, T. R., & Lind, E. A. (1992). A relational model of authority in

groups. In M. P. Zanna (Ed.), Advances in experimental social psychol-

ogy (Vol. 25, pp. 115–191). San Diego, CA: Academic Press.

Vail, K. E., Arndt, J., Motyl, M., & Pyszczynski, T. (2009). Compassionate

values and presidential politics: Mortality salience, compassionate val-

ues, and support for Barack Obama and John McCain in the 2008

presidential election. Analyses of Social Issues and Public Policy, 9,

255–268. doi:10.1111/j.1530-2415.2009.01190.x

van den Bos, K. (2001). Uncertainty management: The influence of un-

certainty salience on reactions to perceived procedural fairness. Journal

of Personality and Social Psychology, 80, 931–941. doi:10.1037/0022-

3514.80.6.931

van den Bos, K., & Lind, E. A. (2002). Uncertainty management by means

of fairness judgments. In M. P. Zanna (Ed.), Advances in experimental

social psychology (Vol. 34, pp. 1–60). San Diego, CA: Academic Press.

Van Vugt, M., & De Cremer, D. (1999). Leadership in social dilemmas:

The effects of group identification on collective actions to provide public

goods. Journal of Personality and Social Psychology, 76, 587–599.

doi:10.1037/0022-3514.76.4.587

Van Vugt, M., Jepson, S. F., Hart, C. M., & De Cremer, D. (2004).

Autocratic leadership in social dilemmas: A threat to group stability.

Journal of Experimental Social Psychology, 40, 1–13. doi:10.1016/

S0022-1031(03)00061-1

Vescio, T. K., Snyder, M., & Butz, D. A. (2003). Power in stereotypically

masculine domains: A social influence strategy 3 stereotype match

model. Journal of Personality and Social Psychology, 85, 1062–1078.

doi:10.1037/0022-3514.85.6.1062

Vroom, V. H., & Yetton, P. W. (1973). Leadership and decisionmaking.

Pittsburgh, PA.: University of Pittsburgh Press.

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-

dation of brief measures of positive and negative affect: The PANAS

scales. Journal of Personality and Social Psychology, 54, 1063–1070.

doi:10.1037/0022-3514.54.6.1063

Weary, G., & Edwards, J. A. (1996). Causal-uncertainty beliefs and related

goal structures. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook

of motivation and cognition: The interpersonal context (Vol. 3, pp.

148–181). New York, NY: Guilford Press.

Weary, G., Tobin, S. J., & Edwards, J. E. (2010). The causal uncertainty

model revisited. In R. M. Arkin, K. C. Oleson, & P. J. Carroll (Eds.),

Handbook of the uncertain self (pp. 78–100). New York, NY: Psychol-

ogy Press.

Weise, D. R., Pyszczynski, T., Cox, C. R., Arndt, J., Greenberg, J.,

Solomon, S., & Kosloff, S. (2008). Interpersonal politics: The role of

terror management and attachment processes in shaping political pref-

erences. Psychological Science, 19, 448 – 455. doi:10.1111/j.1467-

9280.2008.02108.x

Wigboldus, D. H. J., Holland, R. W., & van Knippenberg, A. (2004).

Single target implicit associations. Unpublished manuscript.

Wilson, A. E., & Ross, M. (2001). From chump to champ: People’s

appraisals of their earlier and present selves. Journal of Personality and

Social Psychology, 80, 572–584. doi:10.1037/0022-3514.80.4.572

Wilson, G. D. (1973). The psychology of conservatism. London, England:

Academic Press.

Wilson, T. D., Centerbar, D. B., Kermer, D. A., & Gilbert, D. T. (2005).

The pleasures of uncertainty: Prolonging positive moods in ways people

do not anticipate. Journal of Personality and Social Psychology, 88,

5–21. doi:10.1037/0022-3514.88.1.5

Yukl, G. A. (1998). Leadership in organizations (4th ed.). Englewood

Cliffs, NJ: Prentice-Hall.

(Appendix follows)

19UNCERTAINTY AND LEADERSHIP PREFERENCES

Appendix

German Implicit Association Test Stimuli Used in Studies 1–4 (Bold), Including TheirEnglish Translations

demokratisch autoritar ANGENEHM UNANGENEHM ERFOLG MISSERFOLGdemocratic authoritarian PLEASANT UNPLEASANT SUCCESS FAILURE

sensibel bestimmt JUWEL LÄUSE SIEG FEHLSCHLAGsensitive decisive JEWEL LICE VICTORY MISCARRIAGE

teamfahig direktiv DUFTE MASERN AUFSTIEG VERSAGENteam-player directive FRAGRANCES MEASLES RISE FAILING

liberal energisch ENGEL SCHLEIM TRIUMPH KRISEliberal resolute ANGEL SLIME TRIUMPH CRISIS

integrativ patriarchalisch MEER POCKEN GEWINN VERLUSTintegrative patriarchic SEA POCKS GAIN LOSS

kumpelhaft dominant DIAMANT DRECK GELINGEN FLOPcompanionable dominant DIAMOND DIRT ACHIEVEMENT FLOP

Received March 31, 2010

Revision received January 31, 2011

Accepted February 7, 2011 n

20 SCHOEL, BLUEMKE, MUELLER, AND STAHLBERG